WO2023146012A1

WO2023146012A1 - Method for providing tactical information for team sport

Info

Publication number: WO2023146012A1
Application number: PCT/KR2022/002152
Authority: WO
Inventors: 윤진성; 김현성
Original assignee: 주식회사 핏투게더
Priority date: 2022-01-27
Filing date: 2022-02-14
Publication date: 2023-08-03
Also published as: KR20230115643A

Abstract

A method for providing tactical information for a team sport, according to an aspect of the present disclosure, may comprise the steps of: acquiring a plurality of player-tracking data sets for a plurality of players; acquiring role assignment information by using the plurality of player tracking data sets; determining the primary role assignment on the basis of a plurality of role assignments; determining at least one irregular role assignment from the plurality of role assignments by considering the primary role assignment; on the basis of remaining role assignment information other than the at least one irregular role assignment and some of the player tracking data sets corresponding to the remaining role assignment information, acquiring a plurality of spatial information sets associated with a target duration for each of a plurality of role indexes; and updating the role assignment information by using the plurality of spatial information sets.

Description

How to provide tactical information for team sports

BACKGROUND OF THE INVENTION [0002] The present disclosure relates generally to the field of sports analysis, but is not limited to, more particularly, to techniques for providing tactical information by analyzing tactical changes in team formations and roles in team sports.

Training in modern team sports is largely divided into physical training and tactical training. While physical training aims to improve the player's basic physical abilities such as strength, endurance, and quickness, tactical training aims to improve the ability to perform tactics to obtain overall or local gains during the game.

In tactical training, practical exercises to master local tactics performed by small groups, response in specific situations such as set-pieces, and overall strategies for team formation are important, but to improve team-level understanding of these Tactics analysis and review are also important elements.

Team formation and player positions are the most basic elements in tactical analysis, which are essential for higher-level tactical analysis. However, from a sports analysis point of view, team sports, especially in fluid sports where players interact in real time, such as soccer or basketball, have high-dimensional characteristics that change very fluidly, so most modern sports teams still have no knowledge of formation or position. Analysis is completely dependent on the intuition of a person such as a video analyst.

One object of the present disclosure is to provide a method for providing tactical information for team sports that performs tactical analysis for team sports from player tracking data.

One task of the present disclosure is to provide a method of providing tactical information for team sports that provides information about team formations and/or player positions.

One task of the present disclosure is to provide a method for performing higher-level tactical analysis using information about team formations and/or player positions.

One object of the present disclosure is to provide a method for providing tactical information for a team sport, which obtains spatial information comprising a sequence of location data of a plurality of roles of a team sport from player tracking data.

One object of the present disclosure is to provide a method for providing tactical information for team sports, which divides a target session for a team sports game into a plurality of formation periods based on the time point when the formation of a team participating in the team sports is changed. is to provide

One object of the present disclosure is to provide a method for providing tactical information for a team sport, identifying a formation of a team participating in the team sport, or roles constituting the formation of the team.

One object of the present disclosure is to provide a method for providing tactical information for team sports, which automatically matches a location information acquisition device randomly assigned to a player participating in the team sports with a player identifier.

One task of the present disclosure is to provide tactical information about team sports, which divides a target period into a plurality of role periods based on a time point when the intended role of at least some of the players of a team participating in the team sport changes. is to provide a way

One object of the present disclosure is to provide a method for providing tactical information for team sports, detecting an irregular situation using assignment information for a plurality of roles generated from player tracking data.

One task of the present disclosure is to determine information on the play style of a team or player participating in team sports using assignment information for a plurality of roles generated from player tracking data, providing tactical information about team sports. is to provide a way

An object of the present disclosure is to provide a method for automatically extracting a highlight video from a team sports game video using information on non-regular situations or switch plays.

However, the problem to be solved by the present disclosure is not limited thereto, and may be expanded in various ways without departing from the spirit and scope of the present disclosure.

According to one aspect of the present disclosure, obtaining a plurality of player tracking data sets for each of a plurality of players, each player tracking data set including the corresponding player's positions at a plurality of time points within a target session step; acquiring a plurality of position distributions for each of the plurality of players, each position distribution being generated based on a corresponding player tracking data set; assigning a plurality of role identifiers to each of the plurality of location distributions; a plurality of player-role assignments for each of the plurality of time points, each assignment reflecting a relationship between the plurality of players and the plurality of role identifiers at a first corresponding time point, and the plurality of location distributions and Obtaining - generated by assigning each of the plurality of role identifiers to the plurality of players based on positions of the plurality of players at the first corresponding time point among the plurality of player tracking data sets; A plurality of role assignments for each of the plurality of viewpoints, each role assignment reflecting a positional relationship between the plurality of role identifiers at a second corresponding viewpoint, and a player-role at the second corresponding viewpoint obtaining an allocation and generated based on the positions of the plurality of players at the second corresponding point in time; and dividing the sequence of the plurality of role assignments into at least two role assignment groups using a changepoint detection algorithm.

According to another aspect of the present disclosure, obtaining a plurality of player tracking data sets for each of a plurality of players, each player tracking data set including the corresponding player's positions at a plurality of time points within a target session. step; acquiring a plurality of position distributions for each of the plurality of players, each position distribution being generated based on a corresponding player tracking data set; assigning a plurality of role identifiers to each of the plurality of location distributions; a plurality of player-role assignments for each of the plurality of time points, each assignment reflecting a relationship between the plurality of players and the plurality of role identifiers at a first corresponding time point, and the plurality of location distributions and Obtaining - generated by assigning each of the plurality of role identifiers to the plurality of players based on positions of the plurality of players at the first corresponding time point among the plurality of player tracking data sets; and acquiring a main player-role assignment from the plurality of player-role assignments.

According to another aspect of the present disclosure, obtaining a plurality of player tracking data sets for a plurality of players, each player tracking data set including a sequence of corresponding player's positional data during a target period; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to the - Obtaining; determining a primary role assignment based on the plurality of role assignments;

determining at least one non-canonical role assignment from the plurality of role assignments in view of the primary role assignment; A plurality of spatial information associated with the target period for each of the plurality of role indexes based on remaining role assignment information excluding the at least one non-regular role assignment and a portion of the player tracking data sets corresponding to the remaining role assignment information obtaining sets; and updating the role assignment information using the plurality of spatial information sets. A method of providing tactical information for team sports including, may be provided.

According to another aspect of the present disclosure, obtaining a plurality of player tracking data sets for a plurality of players, each player tracking data set including a sequence of the corresponding player's positional data during a target session; By applying a change point detection algorithm to feature values for each of a plurality of time points generated based on the plurality of player tracking data sets, the time point at which the formation of a team participating in the target session is changed within the target session is reflected. detecting a point of change; and at least two time periods in which a formation of a team participating in the target session is a first team formation, and a first time period in which a formation of the team is a first team formation. dividing into a second team formation comprising a second time period; A method of providing tactical information for team sports including, may be provided.

According to another aspect of the present disclosure, obtaining a plurality of player tracking data sets for a plurality of players, each player tracking data set including a sequence of the corresponding player's positional data during a target session; and a formation structure for the target session generated based on the plurality of player tracking data sets, wherein the formation structure reflects information for determining a formation of a team participating in a team sports game during the target session. A method of providing tactical information for team sports, comprising the step of identifying the formation of the team by the method may be provided.

According to another aspect of the present disclosure, a plurality of player tracking data sets for a plurality of players, each player tracking data set including a sequence of corresponding player's positional data during a target period, is used to determine a plurality of roles. obtaining a plurality of spatial information sets related to the target period for each of the spatial information sets, each spatial information set including a sequence of positional data of a corresponding role during the target period; role space arrangement information - the role space arrangement information includes a plurality of role space arrangements for each of a plurality of viewpoints within the target period, each role space arrangement comprising: a position between the plurality of roles at a first corresponding viewpoint; reflecting the enemy relationship - acquiring; obtaining a formation structure for the target period based on the plurality of role space arrangements, wherein the formation structure reflects information for determining a formation of a team participating in a team sports game during the target period; and determining a formation of the team based on the formation structure. A method of providing tactical information for team sports including, may be provided.

According to another aspect of the present disclosure, obtaining a plurality of player tracking data sets for a plurality of players, each player tracking data set including a sequence of corresponding player's positional data during a target period; A time point at which the role of at least some of the plurality of players is changed within the target period by applying a change point detection algorithm to feature values for each of a plurality of time points generated based on the plurality of player tracking data sets. detecting a point of change that is reflected; and the target period based on the point of change is defined as at least two time periods, the at least two time periods including a first role time period before the point of change and a second role time period after the point of change. Dividing into; A method of providing tactical information for team sports including, may be provided.

According to another aspect of the present disclosure, obtaining a plurality of player tracking data sets for a plurality of players, each player tracking data set including a sequence of corresponding player's positional data during a target period; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to the - Obtaining; and detecting an irregular situation in the target period based on the plurality of role assignments, wherein a difference value between a role assignment at a time point corresponding to the irregular situation and a main role assignment is greater than or equal to a predetermined threshold value. A method of providing tactical information for team sports including, may be provided.

According to another aspect of the present disclosure, obtaining a plurality of player tracking data sets for a plurality of players, each player tracking data set including a sequence of corresponding player's positional data during a target period; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to the - Obtaining; and generating information about a play style of at least one team or at least one player participating in a team sports game based on the plurality of role assignments. A method of providing tactical information for team sports including, may be provided.

According to embodiments of the present disclosure, tactical analysis for team sports may be performed from player tracking data.

According to an embodiment of the present disclosure, spatial information including a sequence of location data of a plurality of roles of a team sport may be obtained from player tracking data.

According to an embodiment of the present disclosure, a target session for a team sports game may be divided into a plurality of formation periods based on a point in time when a formation of a team participating in the team sports is changed.

According to an embodiment of the present disclosure, it is possible to detect a change point in a team formation from player tracking data, search a point where a team formation changes during a game, and identify a team formation for each period.

According to an embodiment of the present disclosure, a formation of a team participating in team sports or roles constituting the formation of the corresponding team may be identified.

According to an embodiment of the present disclosure, a device for obtaining location information randomly assigned to a player participating in team sports may be automatically matched with a player identifier.

According to an embodiment of the present description, the target period may be divided into a plurality of role periods based on a point in time when the indicated roles of at least some of the team players participating in team sports are changed.

According to an embodiment of the present description, a change in a player's position can be identified from player tracking data, and a play style of a team can be extracted.

According to an embodiment of the present disclosure, a role switch may be identified from player tracking data, and a set-piece situation may be retrieved.

According to an embodiment of the present disclosure, an irregular situation may be detected using assignment information for a plurality of roles generated from player tracking data.

According to an embodiment of the present disclosure, information on a play style of a team or player participating in team sports may be determined using assignment information for a plurality of roles generated from player tracking data.

According to an embodiment of the present disclosure, a highlight image may be automatically extracted from a team sports game image using information on an irregular situation or switch play.

The problems, solutions, and effects to be solved by the present invention are not limited to the above-mentioned matters, and matters not mentioned are clearly apparent to those skilled in the art from this specification and the accompanying drawings. You will be able to understand.

1 is an example of a timeline of formation and role changes according to the results of SportsCPD according to an embodiment of the present disclosure.

2 and 3 are graphs of role adjacency according to an embodiment of the present disclosure.

4 relates to an example formation group according to an embodiment of the present disclosure.

5 is a diagram related to formation prediction according to an embodiment of the present disclosure.

6 is a graph of role substitution sequences for each formation period according to an embodiment of the present disclosure.

7 is a diagram related to an example of tactic analysis using a role substitution sequence according to an embodiment of the present disclosure.

8 is a diagram illustrating a relationship between a switch rate and set-piece occurrence according to an embodiment of the present disclosure.

9 is a block diagram showing the configuration of a system for providing tactical analysis of team sports according to an embodiment of the present disclosure.

10-17 are flow diagrams of examples of methods for providing tactical information for team sports according to embodiments of the present disclosure.

18 illustrates an exemplary process for obtaining player tracking data that may be used by embodiments of the present disclosure.

19 is a schematic flowchart of a process for obtaining role assignment information according to an embodiment of the present disclosure.

Fig. 20 is a detailed flow diagram of the non-canonical role assignment process of Fig. 19;

21 is an example of an initial role assignment table based on player tracking data.

22 shows an exemplary initial role assignment result.

23 illustrates role assignment considering the distance between player positions and role positions.

24A to 24C show changes in role assignment of each of a plurality of viewpoints.

25 illustrates an example of a changed role at a time point of role assignment change.

26 is an example of overall role assignment including a role assignment change section.

27 illustrates updating of a spatial information set for each role through role assignment.

28 is an example of a role assignment table showing update of the spatial information set of FIG. 27;

29 is an example of a role assignment table for determining non-canonical role assignment.

FIG. 30 is an example of determining a switch rate according to the role assignment table of FIG. 29 .

31 illustrates updating of spatial information sets for respective roles according to removal of irregular frames.

FIG. 32 exemplarily shows that role assignment is changed at the same point in time according to the update of the spatial information set of FIG. 31 .

33 is a schematic flow diagram of a formation structure acquisition process according to an embodiment of the present disclosure.

34 is a schematic flow diagram of a formation time period segmentation process according to one embodiment of the present disclosure.

FIG. 35 is a detailed flowchart of obtaining spatial arrangement information for the changing point detection process of FIG. 34;

Fig. 36 is a detailed flow chart of obtaining distance information for the point of change detection process in Fig. 34;

37 is a schematic flow diagram of a matching process for formation identification, position identification and player identifier according to one embodiment of the present disclosure.

38 illustrates Delaunay triangulation for role adjacency determination.

39 is an example of a role adjacency matrix representing a positional relationship between roles.

40 is an exemplary diagram of a process of dividing a formation time period based on mutual distances of each of a plurality of frames according to an embodiment of the present disclosure.

41 is an illustration of a process for dividing a formation time period based on a sequence of distances between role adjacency matrices of each of adjacent frames according to an embodiment of the present disclosure.

42 shows an exemplary label of training data for formation identification based on an artificial neural network according to an embodiment of the present disclosure.

43 is a conceptual diagram of a role alignment process for clustering between different formation periods or generating training data.

44 is a schematic flow diagram for a role time period segmentation process according to an embodiment of the present disclosure.

45 is a detailed flow chart of distance information acquisition for the point of change detection process in FIG. 44;

46 is an exemplary view of sequence acquisition of distances between role assignments between adjacent frames for role period division.

47 is an exemplary view of role segment division using a change point detection algorithm.

48 is a schematic flowchart of a process of detecting an irregular situation and extracting highlight image data according to an embodiment of the present disclosure.

49 is a schematic flowchart of a play style analysis and highlight image data extraction process according to an embodiment of the present disclosure.

50 is a detailed flowchart of the play style information generation process in FIG. 49;

51 is an example of frequency analysis of role assignment for generating play style information.

The embodiments described in this description are intended to clearly explain the spirit of the present invention to those skilled in the art to which the present invention belongs, and the present invention is not limited by the examples described in this description, and the present invention The scope of should be construed as including modifications or variations that do not depart from the spirit of the present invention.

The terminology used in this description is a selection of general terms that are currently widely used as much as possible in the technical field to which the present invention belongs, but these terms are selected according to the intention, custom, or the emergence of new technologies of those skilled in the art in the technical field to which the present invention belongs. meaning may vary. However, in the case where a specific term is defined and used in an arbitrary meaning, the meaning of the term will be separately described. Therefore, the terms used in this description should be interpreted based on the actual meaning of the term and the overall content of this description, not the simple name of the term.

The drawings attached to this description are for easily explaining the present invention, and the shapes shown in the drawings may be exaggerated or abbreviated as necessary to help the understanding of the present invention, so the present invention is not limited by the drawings. .

In the present description, if it is determined that a detailed description of a known configuration or function related to the present invention may obscure the gist of the present invention, a detailed description thereof will be omitted if necessary.

Here, the method further comprises: obtaining a primary player-role assignment from the plurality of player-role assignments; and determining a non-regular player-role assignment of at least one of the plurality of player-role assignments based on the main player-role assignment, wherein the sequence of plurality of role assignments comprises: Among the batches, only the remainder except for the at least one non-regular player-role assignment may be included.

Here, the method comprises: assigning the target session to at least two time periods corresponding to each of the at least two role placement groups - the at least two time periods, wherein the main formation of a team participating in the target session is a first team and a second time period in which the primary formation of the team is a second team formation.

wherein the method further comprises determining the first team formation from at least one first role assignment with respect to the first time period and the second team formation from at least one second role placement with respect to the second time period. Step; may further include.

Here, the main player-role assignment may be the most frequent player-role assignment among the plurality of player-role assignments.

Here, the method may further include determining a plurality of positions for each of the plurality of players based on the main player-role assignment.

Here, the method may further include selecting a secondary player-role assignment from the plurality of player-role assignments.

Here, the secondary player-role assignment may be the second most frequent player-role assignment among the plurality of player-role assignments.

Here, the method may further include acquiring information on position conversion by comparing the main player-role assignment and the secondary player-role assignment.

Here, the method may further include acquiring switch rate information regarding the plurality of players-role assignments based on the main player-role assignments.

Here, the method may further include detecting an irregular situation in the target session based on the switch rate information.

A method for providing tactical information for a team sport according to one aspect of the present disclosure includes a plurality of player tracking data sets for a plurality of players, each player tracking data set having a corresponding player's positional data for a target period. comprising a sequence; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to the - Obtaining; determining a primary role assignment based on the plurality of role assignments; determining at least one non-canonical role assignment from the plurality of role assignments in view of the primary role assignment; A plurality of spatial information associated with the target period for each of the plurality of role indexes based on remaining role assignment information excluding the at least one non-regular role assignment and a portion of the player tracking data sets corresponding to the remaining role assignment information obtaining sets; and updating the role assignment information using the plurality of spatial information sets. can include

Here, the main role assignment may be the most frequent role assignment among the plurality of role assignments.

Here, the step of determining the non-canonical role assignment comprises: information about distances from the primary role assignment for each of the plurality of role assignments, wherein the distance is between each of the plurality of role assignments and the primary role assignment. Obtaining - reflecting different degrees from each other; and determining the non-canonical role assignment based on the information about the distances.

Here, the distance may reflect a switch rate from the main role assignment for each of the plurality of role assignments.

The method may further include determining whether an additional update is required for the plurality of spatial information sets or the role assignment information; may further include.

Here, the step of determining whether the additional update is required may determine whether the additional update is required based on whether at least a part of the role assignment information is changed by the step of updating the role assignment information. .

Here, the step of determining whether the additional update is required may include determining whether the additional update is required based on whether the time points at which the role assignment of each time point is changed by the step of updating the role assignment information are equal to or less than a predetermined threshold value. can decide whether

Here, the step of determining whether the additional update is required may determine whether the additional update is required based on whether a predetermined number of updates is achieved.

Here, the method comprises dividing a target session for a game of a team sport into at least two time periods, wherein the at least two time periods change the formation of a team participating in the game or the role of at least some of the plurality of players. comprising a first period of time and a second period of time divided on the basis of change; wherein the target period may be either the first period of time or the second period of time.

An apparatus for providing tactical information for a team sport according to an aspect of the present disclosure includes a processor and a memory, wherein the processor includes a plurality of player tracking data sets for a plurality of players, each player tracking data set comprising a target obtains - including a sequence of positional data of the corresponding player over the period; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to - Obtain; determine a primary role assignment based on the plurality of role assignments; determine at least one non-canonical role assignment from the plurality of role assignments in view of the primary role assignment; A plurality of spatial information associated with the target period for each of the plurality of role indexes based on remaining role assignment information excluding the at least one non-regular role assignment and a portion of the player tracking data sets corresponding to the remaining role assignment information get sets; and updating the role assignment information using the plurality of spatial information sets; can be configured to

According to one aspect of the present disclosure, a non-transitory computer readable storage medium storing instructions executable by a processor, wherein the instructions are executed by the processor to cause the processor to: obtain player tracking data sets, each player tracking data set comprising a sequence of corresponding player's positional data during a target period; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to - Obtain; determine a primary role assignment based on the plurality of role assignments; determine at least one non-canonical role assignment from the plurality of role assignments in view of the primary role assignment; A plurality of spatial information associated with the target period for each of the plurality of role indexes based on remaining role assignment information excluding the at least one non-regular role assignment and a portion of the player tracking data sets corresponding to the remaining role assignment information get sets; and updating the role assignment information using the plurality of spatial information sets; can be configured to do so.

A method of providing tactical information for a team sport in accordance with one aspect of the present disclosure includes a plurality of player tracking data sets for a plurality of players, each player tracking data set having a corresponding player's location data during a target session. comprising a sequence; By applying a change point detection algorithm to feature values for each of a plurality of time points generated based on the plurality of player tracking data sets, the time point at which the formation of a team participating in the target session is changed within the target session is reflected. detecting a point of change; and at least two time periods in which a formation of a team participating in the target session is a first team formation, and a first time period in which a formation of the team is a first team formation. dividing into a second team formation comprising a second time period; can include

Here, each characteristic value for each of the plurality of viewpoints includes positions of the plurality of players at each viewpoint; a player space arrangement that reflects the positional relationship between the plurality of players at each viewpoint; positions of the plurality of roles at each point in time; or a role space arrangement that reflects a positional relationship between the plurality of roles at each viewpoint; It may be generated based on at least one of

Here, the step of detecting the change point includes role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of points in time in the target session, and each the role assignment indicates a plurality of role indices assigned to the plurality of players at a first corresponding point in time; obtaining a plurality of spatial information sets associated with the target session for each of the plurality of role indexes based on the role assignment information and the player tracking data sets; and role space arrangement information including a plurality of role space arrangements for each of the plurality of viewpoints, each role space arrangement reflecting a positional relationship between the plurality of role indices at a second corresponding viewpoint. obtaining; can include

Here, the player space arrangement at each time point includes player adjacency information reflecting whether or not a first player among the plurality of players is adjacent to a second player among the plurality of players at each time point, The role space arrangement at each time point may include role adjacency information reflecting whether a first role among the plurality of roles is adjacent to a second role among the plurality of roles at each time point.

Here, the player adjacency information includes a player adjacency matrix reflecting whether each of the plurality of players is adjacent to each other with respect to each of the plurality of players, and the role adjacency information includes each of the plurality of roles. and a role adjacency matrix reflecting whether each of the plurality of roles is adjacent to each other.

Here, the player adjacency matrix is obtained based on performing Delaunay triangulation on the plurality of players, and the role adjacency matrix performs Delaunay triangulation on the plurality of roles can be obtained based on

Here, the step of detecting the change point may include information about a distance between a feature value of a first point in time and a feature value of a second point in time among feature values for each of the plurality of points in time - the distance between the feature values Obtaining ?, which reflects different degrees of ?; and applying the changing point detection algorithm to the information about the distance.

Here, the information on the distance may include an inter-frame distance matrix reflecting a distance for each of the feature values of the plurality of views from each of the feature values of the plurality of views.

Here, the distance-related information may include distances for each of the feature values of the plurality of viewpoints from the reference characteristic value of the reference viewpoint.

Here, the information on the distance is time-series distances reflecting a distance between a feature value for a target viewpoint and a feature value for a viewpoint previous to the target viewpoint, wherein the time-series distance is calculated for each of a plurality of viewpoints. Can contain sequences.

Here, the change point detection algorithm may include a discrete g-segmentation algorithm.

Here, the method includes determining whether the change point is valid based on characteristics of at least one divided period; may further include.

Here, the step of determining whether the change point is valid may include determining that the change point is valid in response to determining that the significance value (Scan Statics p) of the scan statistic according to the change point is equal to or less than a predetermined first threshold value. can

Here, the step of determining whether the point of change is valid may include determining that the point of change is valid in response to a determination that the time lengths of both the first time period and the second time period are equal to or greater than a second predetermined threshold. can

Here, the step of determining whether the change point is valid may include determining an average value of feature values for a plurality of points of time included in the first time period and feature values of a plurality of points of time included in the second time period. In response to a determination that the difference in mean value is equal to or greater than a third predetermined threshold value, it may be determined that the point of change is valid.

An apparatus for providing tactical information for a team sport according to an aspect of the present disclosure includes a processor and a memory, wherein the processor includes a plurality of player tracking data sets for a plurality of players, each player tracking data set comprising a target obtains - containing a sequence of corresponding player's positional data during the session; By applying a change point detection algorithm to feature values for each of a plurality of time points generated based on the plurality of player tracking data sets, the time point at which the formation of a team participating in the target session is changed within the target session is reflected. detecting a change point; and at least two time periods in which a formation of a team participating in the target session is a first team formation and a first time period in which a formation of the team participates in the target session is a first team formation. 2nd Team Formation, including the 2nd Time Period - Split by; can be configured to

According to one aspect of the present disclosure, a computer readable storage medium storing instructions executable by a processor, wherein the instructions are executed by the processor to cause the processor to: obtaining sets, each player tracking data set containing a sequence of positional data of the corresponding player during the target session; By applying a change point detection algorithm to feature values for each of a plurality of time points generated based on the plurality of player tracking data sets, the time point at which the formation of a team participating in the target session is changed within the target session is reflected. detecting a change point; and at least two time periods in which a formation of a team participating in the target session is a first team formation and a first time period in which a formation of the team participates in the target session is a first team formation. 2nd Team Formation, including the 2nd Time Period - Split by; can be configured to do so.

A method of providing tactical information for a team sport in accordance with one aspect of the present disclosure includes a plurality of player tracking data sets for a plurality of players, each player tracking data set having a corresponding player's location data during a target session. comprising a sequence; and a formation structure for the target session generated based on the plurality of player tracking data sets, wherein the formation structure reflects information for determining a formation of a team participating in a team sports game during the target session. It may include identifying the formation of the team as.

Here, the method comprises dividing the target session into at least two time periods, wherein the at least two time periods include a first time period and a second time period divided based on a formation change of a team participating in the game. - dividing by , wherein identifying the formation comprises: identifying a first formation of the team for the first period of time based on a first formation structure for the first period of time; and identifying a second formation of the team for the second period of time based on the second formation structure for the second period of time; can include

In the step of dividing the target session, the formation of the team within the target session is determined by applying a change point detection algorithm to feature values for each of a plurality of time points generated based on the plurality of player tracking data sets. detecting a change point reflecting a changed time point; Dividing the target session into the at least two time periods based on the change point.

Here, the formation structure for the target session includes sequences of positions during the target session for each of the plurality of players; average positions during the target session for each of the plurality of players; a sequence of player space arrangements for each viewpoint during the target session, each player space arrangement reflecting a positional relationship between the plurality of players at each viewpoint; or an average of player space arrangements for each time point during the target session; It may be generated based on at least one of

Here, the formation structure for the target session includes sequences of positions during the target session for each of a plurality of roles; average positions during the target session for each of the plurality of roles; a sequence of role space arrangements for each viewpoint during the target session, each role space arrangement reflecting a positional relationship among the plurality of roles of each viewpoint; or an average of role space placements for each point in time during the target session; It may be generated based on at least one of

Here, the step of identifying the formation includes determining a formation identifier corresponding to the formation structure using an artificial neural network-based formation identification model, wherein the formation identification model includes a plurality of training data sets - each training The data set may be generated by training an artificial neural network based on formation structure samples and labeled formation identifiers corresponding to the formation structure samples.

Here, the formation structure sample may include a plurality of normalized data elements arranged by assigning each of a plurality of data elements included in the formation structure sample to each of a plurality of reference data elements included in the reference formation structure. can

Wherein, identifying the formation comprises: acquiring a formation identifier corresponding to the first formation structure group by matching the formation structure to a first formation structure group among a plurality of formation structure groups; The plurality of formation structure groups may be generated by clustering a plurality of formation structure samples and labeling each cluster with a formation structure identifier.

Here, each formation structure sample may include a plurality of normalized data elements arranged by assigning each of a plurality of data elements included in the formation structure sample to each of a plurality of reference data elements included in the reference formation structure. can

Here, the method includes obtaining formation data corresponding to the identified formation, wherein the formation data includes information on a plurality of positions for the formation and position characteristic values for each of the plurality of positions. step; and player-position assignments for each of a plurality of players, wherein each assignment is based on position characteristic values for each of the plurality of players and position characteristic values for each of the plurality of positions. Obtaining - generated by assigning each of the plurality of positions to each of the plurality of positions; may further include.

Here, each location characteristic value is a representative location for each player or each position; or distribution of positions for each player or each position; It may be generated based on at least one of

Here, the method includes player information of a team participating in the target session - the player information includes a plurality of positions for the formation of the team and player identifiers corresponding to each of the plurality of positions. - obtaining ; and matching each of the plurality of players to each of the plurality of player identifiers based on the player-position assignments.

Here, each player identifier may include at least one of a player name and a uniform number.

A method for providing tactical information for a team sport according to one aspect of the present disclosure includes a plurality of player tracking data sets for a plurality of players, each player tracking data set having a corresponding player's positional data for a target period. a plurality of spatial information sets related to the target period for each of a plurality of roles based on a sequence, each spatial information set including a sequence of position data of a corresponding role during the target period; obtaining; role space arrangement information - the role space arrangement information includes a plurality of role space arrangements for each of a plurality of viewpoints within the target period, each role space arrangement comprising: a position between the plurality of roles at a first corresponding viewpoint; reflecting the enemy relationship - acquiring; obtaining a formation structure for the target period based on the plurality of role space arrangements, wherein the formation structure reflects information for determining a formation of a team participating in a team sports game during the target period; and determining a formation of the team based on the formation structure. can include

Here, each of the role space arrangements includes role adjacency information reflecting whether a first role among the plurality of roles is adjacent to a second role among the plurality of roles at the first corresponding time point. can do.

Here, the role adjacency information may include a role adjacency matrix reflecting whether each of the plurality of roles is adjacent to each other with respect to each of the plurality of roles.

Here, the role adjacency matrix may be obtained based on performing Delaunay triangulation on the plurality of roles.

Here, the formation structure may be a sequence of the plurality of role adjacency matrices for each of the plurality of viewpoints.

Here, the formation structure may be an average role adjacency matrix reflecting an average of the plurality of role adjacency matrices for each of the plurality of viewpoints.

A method for providing tactical information for a team sport according to one aspect of the present disclosure includes a plurality of player tracking data sets for a plurality of players, each player tracking data set having a corresponding player's positional data for a target period. comprising a sequence; A time point at which the role of at least some of the plurality of players is changed within the target period by applying a change point detection algorithm to feature values for each of a plurality of time points generated based on the plurality of player tracking data sets. detecting a point of change that is reflected; and the target period based on the point of change is defined as at least two time periods, the at least two time periods including a first role time period before the point of change and a second role time period after the point of change. Dividing into; can include

Here, each feature value for each of the plurality of views may include role assignment of each view, wherein the role assignment indicates a plurality of role indices assigned to the plurality of players at a corresponding view point. there is.

Here, in the step of detecting the change point, role assignment information, wherein the role assignment information includes a plurality of role assignments for each of a plurality of points in time within the target period, is obtained by using the plurality of player tracking data sets. Acquisition may be further included.

Here, the distance between the first role assignment and the second role assignment may reflect the number of different roles between the first role assignment and the second role assignment.

In the detecting of the change point, the change point may be detected based on role assignments other than at least one non-regular role assignment among a plurality of role assignments for each of the plurality of viewpoints.

Here, the non-regular role assignment may include determining a main role assignment based on the plurality of role assignments; and determining at least one irregular role assignment from the plurality of role assignments in consideration of the main role assignment.

Here, the step of determining the non-canonical role assignment comprises: information about distances from the primary role assignment for each of the plurality of role assignments, wherein the distance is between each of the plurality of role assignments and the primary role assignment. Obtaining - reflecting different degrees from each other; and determining the irregular role assignment based on the information about the distances.

Here, the step of determining whether the point of change is valid determines that the point of change is valid in response to a determination that the time lengths of both the first role time period and the second role time period are greater than or equal to a second predetermined threshold. can judge

Here, the method comprises dividing a target session for a game of team sports into at least two formation time periods, wherein the at least two formation time periods are divided based on a change in formation of a team participating in the game; a first formation time period; and a second formation time period, wherein the target period may be either a first formation time period or a second formation time period.

An apparatus for providing tactical information for a team sport according to an aspect of the present disclosure includes a processor and a memory, wherein the processor includes a plurality of player tracking data sets for a plurality of players, each player tracking data set comprising a target obtains - including a sequence of positional data of the corresponding player over the period; A time point at which the role of at least some of the plurality of players is changed within the target period by applying a change point detection algorithm to feature values for each of a plurality of time points generated based on the plurality of player tracking data sets. detecting a point of change that reflects; and based on the point of change, the target period is defined as at least two time periods, wherein the at least two time periods include a first role time period before the point of change and a second role time period after the point of change. split into; can be configured to

According to one aspect of the present disclosure, a computer readable storage medium storing instructions executable by a processor, wherein the instructions are executed by the processor to cause the processor to: obtaining sets, each player tracking data set containing a sequence of corresponding player's positional data during a target period; A time point at which the role of at least some of the plurality of players is changed within the target period by applying a change point detection algorithm to feature values for each of a plurality of time points generated based on the plurality of player tracking data sets. detecting a point of change that reflects; and based on the point of change, the target period is defined as at least two time periods, wherein the at least two time periods include a first role time period before the point of change and a second role time period after the point of change. split into; can be configured to do so.

A method for providing tactical information for a team sport according to one aspect of the present disclosure includes a plurality of player tracking data sets for a plurality of players, each player tracking data set having a corresponding player's positional data for a target period. comprising a sequence; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to the - Obtaining; and detecting an irregular situation in the target period based on the plurality of role assignments, wherein a difference value between a role assignment at a time point corresponding to the irregular situation and a main role assignment is greater than or equal to a predetermined threshold value. can include

Here, the difference between the role assignment at the time corresponding to the non-regular situation and the main role assignment may reflect a switch rate between the role assignment at the time corresponding to the non-regular situation and the main role assignment.

Here, the method includes obtaining full-time image data for the target period; and extracting highlight image data including image data of a viewpoint corresponding to the non-regular situation from the full-time image data. may further include.

Here, the method divides a target session for a team sports game into at least two role time periods, wherein the at least two role time periods are divided based on a point in time when the role of at least some of the plurality of players is changed. comprising a first role time period and a second role time period, wherein the target period may be either the first role time period or the second role time period.

An apparatus for providing tactical information for a team sport according to an aspect of the present disclosure includes a processor and a memory, wherein the processor includes a plurality of player tracking data sets for a plurality of players, each player tracking data set comprising a target obtains - including a sequence of positional data of the corresponding player over the period; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to - Obtain; and detecting an irregular situation within the target period based on the plurality of role assignments, wherein a difference value between a role assignment at a time point corresponding to the irregular situation and a main role assignment is greater than or equal to a predetermined threshold value.

According to one aspect of the present disclosure, a computer readable storage medium storing instructions executable by a processor, wherein the instructions are executed by the processor to cause the processor to: obtaining sets, each player tracking data set containing a sequence of corresponding player's positional data during a target period; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to - Obtain; and detecting an irregular situation in the target period based on the plurality of role assignments, wherein a difference value between a role assignment at a time point corresponding to the irregular situation and a main role assignment is equal to or greater than a predetermined threshold value; can be configured to do so.

A method for providing tactical information for a team sport according to one aspect of the present disclosure includes a plurality of player tracking data sets for a plurality of players, each player tracking data set having a corresponding player's positional data for a target period. comprising a sequence; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to the - Obtaining; and generating information about a play style of at least one team or at least one player participating in a team sports game based on the plurality of role assignments. can include

Here, the information on the play style may include information about a time length of switch play reflecting a temporary role change by at least some of the plurality of players; information about the number of times of the switch play; Alternatively, at least one of information about roles in which the switch play occurred may be included.

Here, the generating of the play style information may include determining a main role assignment based on the plurality of role assignments; determining a sub-role assignment based on the plurality of role assignments; and obtaining information about the switch play by comparing the main role assignment with the sub-role assignment.

Here, the secondary role assignment may be the second most frequent role assignment among the plurality of role assignments.

Here, the switch play includes a first switch play reflecting role alternation between roles corresponding to the first role group and a second switch play reflecting role alternation between roles corresponding to the second role group, At least one of the plurality of viewpoints may be reflected in both the first switch play and the second switch play.

Obtaining full-time image data for the target period; and extracting highlight video data including video data for a time point corresponding to the switch play from the full-time video data. may further include.

An apparatus for providing tactical information for a team sport according to an aspect of the present disclosure includes a processor and a memory, wherein the processor includes a plurality of player tracking data sets for a plurality of players, each player tracking data set comprising a target obtains - including a sequence of positional data of the corresponding player over the period; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to - Obtain; and generating information about a play style of at least one team or at least one player participating in a team sports match based on the plurality of role assignments; can be configured to

According to one aspect of the present disclosure, a computer readable storage medium storing instructions executable by a processor, wherein the instructions are executed by the processor to cause the processor to: obtaining sets, each player tracking data set containing a sequence of corresponding player's positional data during a target period; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to - Obtain; and generating information about a play style of at least one team or at least one player participating in a team sports match based on the plurality of role assignments; can be configured to do so.

1. 서설1. Introduction

In fluid team sports such as football and basketball, analyzing team formations is one of the most intuitive ways to understand tactics from the practitioner's point of view. However, existing approaches have limitations in not fully reflecting the actual game situation even if they assume that the team formation is consistent throughout the game or assign the formation for each scene. With this problem in mind, in this article, SportsCPD (Sports Change-Point Detection), a change point detection framework that distinguishes tactically intended changes in formations and player positions during team sports games from temporary changes using spatio-temporal tracking data, is developed. Suggest.

Here, the term "player position" can be interpreted as referring to a role assigned to a player in team sports (eg, center forward or central defender in soccer), and the term "team formation" refers to a team's affiliation. Some form of combination of roles assigned to all or most of the players (e.g. field players excluding the goalkeeper in soccer) (e.g. "4-4-2" or "3-4-3" in soccer). ) can be interpreted as referring to

Here, as an example of "intentional change", a team that used a "4-4-2" formation in the first half of a soccer match adopts a "3-4-3" formation in the second half under the tactical judgment of the coaches. An example of a "temporary change" is a temporary change, such as playing the role of an attacker for a short period of time by overlapping a defender in a soccer game. , random or temporary occurrence of such changes.

SportsCPD according to an embodiment of the present disclosure may detect a point at which a team formation changes (hereinafter referred to as a “formation change point”). Hereinafter, this is referred to as Formation Change-Point Detection (FormCPD). Illustratively, FormCPD first acquires information reflecting the roles assigned to players for each scene, obtains information reflecting the positional relationship between the roles, and then obtains information about the positional relationship between the roles. This can be done by detecting formation change points from the sequence.

In addition, SportsCPD according to an embodiment of the present disclosure can detect a player position, that is, a point at which a role assigned to a player changes (hereinafter referred to as a “role change point”). Hereinafter, this is referred to as Role Change-Point Detection (RoleCPD). Illustratively, RoleCPD may be performed by detecting a role change point from a sequence of information reflecting roles assigned to players.

Here, information reflecting roles assigned to players may be referred to as "player-role assignments". For example, player-role assignment can be expressed as a role permutation, which will be described in more detail later.

Also, here, information reflecting the positional relationship between roles may be referred to as "role arrangement". For example, the role arrangement may be expressed as a role-adjacency matrix, which will be described in detail later.

According to one example, detection of change points may be performed using a nonparametric change point detection technique. However, the change point detection technique is not limited thereto.

According to the result of evaluating the method according to the embodiment of the present disclosure using actual measurement data recorded by workers in the field, the method according to the embodiment of the present disclosure can accurately detect tactical change points and predict the formation of each segment. Confirmed. In addition, in the embodiments of the present description, practical examples that can be interpreted and utilized by those skilled in the art will be mentioned.

2. 개요2. Overview

In fluid team sports such as soccer or basketball, players interact with each other while maintaining their roles within a specific formation. Therefore, analyzing team formations is one of the most intuitive ways to understand tactics from the point of view of practitioners in the field. Since it is clear that team formation has a significant effect on player positions during a game, there have been previous studies in the field of sports science to estimate team formation during a game using spatio-temporal tracking data.

However, the results of previous approaches differed from formation changes in actual matches. Many approaches assume that team formations are consistent throughout a game or session, but this does not reflect real-life situations in which coaches often change team formations in response to various in-game situations during a game. Other approaches have considered changing formations, but have assigned formations for each scene or for each offensive/defensive period. When this allocation method is used, considering that a coach usually instructs only a few changes during a game, there is a problem in that team formation changes too frequently, contrary to the coach's intention.

To fill this gap, an embodiment of the present disclosure proposes a changepoint detection framework called SportsCPD that distinguishes tactically intended changes in formation and role from temporary changes in team sports games.

SportsCPD may be exemplarily performed as follows.

First, roles are assigned to players for each scene. Illustratively, role assignment to players for each scene may be performed as follows. In the assignment of roles to players described later, the concept of role representation presented in Balkowski's thesis "Large-scale analysis of soccer matches using spatiotemporal tracking data (2014)" was partially borrowed.

Player tracking data sets can be acquired. Each player tracking data set relates to the location of an individual player, and may include information about the locations of the corresponding player within a session to be analyzed (hereinafter referred to as “target session”).

Here, the player tracking data set may be obtained, for example, through a positioning sensor (eg, a GPS sensor or an LPS sensor) worn and/or attached to the player or through image analysis of a sports video, and the stadium It may be expressed according to a field coordinate system. The stadium coordinate system may refer to a coordinate system having a longitudinal direction and a width direction of the stadium as axes with a point (for example, a center or a corner point of the stadium) as an origin in a field where a sports game is held. In addition, the player tracking data set may include positions in a time-series form. In this case, the time interval may be the same as the sampling interval of the sensor or image analysis, but this is not necessarily the case.

Roles can be assigned to each player based on the player tracking data set. Here, role assignment may be performed by assigning a role identifier to a player. First, under the assumption that the players have a certain role throughout the session, each of the player's position distributions during the session obtained from the player tracking data sets can be determined as the initial position distributions of the roles over the entire session. Now, we can assign roles to players according to their positions, given the initial position distribution of roles. As a simple example, roles can be assigned to players on a scene-by-scene basis such that the sum of the distances between the average position of the initial distribution of positions of roles assigned to an individual player over the players and the position of that individual player in the scene is minimized. there is. However, instead of simply minimizing the sum of distances, various techniques may be used. For example, a Hungarian algorithm such as a cost matrix based on log probability density may be used. Next, the location distribution of the roles can be newly created considering the roles assigned to the players for each scene. The initial position distribution is assumed to have a player in only one role throughout the session, so it is only built from positions pertaining to one player over the entire session, while the newly constructed position distribution is based on the player performing one role scene-by-scene. Since it can be different, it can be created from the position of a different player for each scene. If the location distributions of the roles are newly created, the roles can be assigned to the players by considering the locations of the players for each scene. Now, if the above operation is repeated a sufficient number of times or until the role allocation or position distribution is sufficiently converged, the final player-role allocation can be obtained.

Meanwhile, the above player-role assignment process may be understood to be effective in a situation where each of the players is playing a specific position on a team formation. Thus, an irregular situation in which the team formation is broken, such as a set-piece situation, may be treated as tainted data in assigning roles. Therefore, it is possible to improve the accuracy of player-role assignment by performing pre-processing to remove non-regular situations among scenes. Specifically, if an irregular situation is detected, player-role assignment may be performed except for this. Specifically, the position distribution of the roles may be calculated using only the player tracking data set of the scenes other than the scene corresponding to the non-regular situation.

Detection of non-canonical situations can be in a variety of ways.

For example, an irregular situation is a situation in which a formation is destroyed. Since a significant number of irregular situations may occur during set-pieces, detection of irregular situations may be performed using characteristics occurring in set-pieces. For example, in soccer, in a set-piece situation such as a free kick, corner kick or penalty kick, the position of the ball is temporarily fixed (and the fixed position is specified in the case of a corner kick or penalty kick), and the players move in a relatively dense positional relationship. none or very small Therefore, if the position of the ball is fixed for a certain period of time or below the threshold (in the case of a corner kick or penalty kick, a condition based on a specific position can be added), the players must be in a fixed position for a certain period of time or below the threshold. When there is movement (in the case of corner kicks or penalty kicks, a condition based on a specific position can be added), and/or when players are concentrated for a certain amount of time (this is illustratively, a predetermined number or more players within a predetermined radius) It can be determined that the scene exists within the distance or the sum of the distances between the players is less than a predetermined distance), the scene can be judged to correspond to an irregular situation.

As another example, an irregular situation may be detected based on the extent to which players deviate from a designated role. Specifically, when roles are assigned to players for each scene, dominant player-role assignment can be determined. Here, the main player-role assignment is a role-assignment mainly used by a team within the target session, that is, a role-assignment according to an indicated team formation. Primary player-role assignments can be determined based on frequency from scene-by-scene player-role assignments in the target session. For example, determining the most frequent player-role assignment as the primary player-role assignment, given that team formations typically dictated in team sports are one or less than a few over target sessions, or a predetermined number of players in order of frequency. Player-role assignments may be determined as primary player-role assignments. Here, when the formation period described later is determined, it may be desirable to assume that there is one main player-role assignment for each formation period, but note that the main player-role assignment does not necessarily have to be one for the entire target session. Should be. When the main player-role assignment is determined, an irregular situation may be detected by comparing the main player-role assignment with the player-role assignment for each scene. For example, when the player-role assignment is expressed as a permutation sequence as described later, if the distance between the permutation sequence of the main player-role assignment and the permutation sequence of the player-role assignment for each scene is greater than a threshold value, the scene corresponds to a non-regular situation. It can be judged that In addition, if there are multiple main player-role assignments, if the minimum distance between the main player-role assignment and the player-role assignment for each scene is greater than or equal to a threshold value, the scene may be determined to correspond to an irregular situation.

As another example, considering that the team formation is suddenly destroyed in an irregular situation, a case where a change in the sequence of player-role assignment for each scene exceeds a certain level may be detected as an irregular situation. Specifically, a case where the distance of the substitution sequence of adjacent player-role assignments is equal to or greater than a threshold value may be determined as an irregular situation.

After assigning roles to players for each scene, change point detection is performed in the following two steps. (1) Formation changing points are detected for each session, and (2) role changing points are detected for each segment obtained from the results. Hereinafter, each step will be described in detail.

At the FormCPD stage, the formation period can be determined in the target session by detecting formation change points using player-role assignment. Here, the formation period may mean a period during which the indicated formation is constant. In addition, formations can be identified by formation period.

For example, in FormCPD, formation change points can be detected from role arrangements.

Here, the role arrangement relates to a positional relationship between role identifiers, and may be generated from player-role assignment and player tracking data sets. In addition, since the role assignment can be generated for each scene from the player-role assignment for each scene and the positions of the players in the scene extracted from the player tracking data set, a sequence of role assignments can be obtained for the target session.

For example, the role arrangement in a specific scene can be expressed as coordinates between roles on the stadium coordinate system, which is the positional coordinates of the players in the scene extracted from the player tracking data set through player-role assignment in the scene can be created by assigning them to roles.

As another example, role placement may calculate an adjacency relationship between roles. Here, the adjacency relation relates to positional adjacency between roles and may be referred to as “role topology”. In the role topological positional relationship, the positional relationship between roles can be determined based on the adjacency between roles based on contextual adjacency instead of actual values such as location on the playing field or distance accordingly. For example, the role phase position relationship is a role-adjacency matrix calculated from Delaunay triangulation with reference to what was presented in Narizuka and Yamazaki's paper "Characterization of the formation structure in team sports (2017)" can be expressed as A role adjacency matrix using this can be obtained as follows. First, positions of roles may be obtained for each scene. In this case, the position may be based on a reference coordinate system. Next, Delaunay triangulation, that is, triangulation is performed so that the minimum angle of all angles is maximized on the polynomial with each role as a vertex, and the roles sharing the triangle are A role adjacency matrix, which is a matrix defining adjacent roles and defining roles that are not adjacent as non-adjacent roles, may be obtained.

Next, considering that the role arrangement is time-series information provided in the form of a sequence for each scene, a change point may be detected from the calculated role arrangement. In this case, it goes without saying that various change point detection algorithms may be used. The change point detection algorithm may include an online change point detection algorithm that detects a change point from an input sequence in real time and an offline change point detection algorithm that detects a change point from a completed sequence. Examples of online algorithms include sequential hypothesis testing, which detects change points by considering factors such as false alarm rate, misdetection rate, detection delay, etc. streaming algorithms, etc., examples of off-line algorithms include change-in-mean detection according to hypothesis verification method, maximum likelihood estimation according to change over time, or anomaly regression method (two -phase regiression) and the like. Therefore, in the present description, change point detection should be understood as a comprehensive concept including any sequence segmentation or statistical analysis.

According to an example, a change point may be detected from a sequence of a role adjacency matrix. In this case, a nonparametric change point detection algorithm may be applied to detect the change point, without limitation, considering that the role adjacency matrix is high-dimensional data having repetitive observations. As a non-parametric algorithm, for example, discrete g-segmentation can be used. Discrete g-segmentation can be performed by referring to Song and Chen's paper "Asymptotic distribution-free change-point detection for data with repeated observations (2020)".

The detected change point may divide the game into several formation periods in which it is assumed that the team formation is maintained within one formation period.

In addition, FormCPD obtains a formation structure for each of the team formation periods, that is, the segments of the target session, and from this, it is possible to identify the formation in the corresponding segment. For example, the formation structure in each segment may be expressed as an average role adjacency matrix during that segment. For another example, the formation structure in each segment may be expressed as position information (eg, average position or position distribution) for each role of a scene corresponding to the main player-role assignment during the segment. In addition, the expressed formation structure can be interpreted in the language of those working in the field (eg, "4-4-2" or "3-4-3"). For example, the formation structure expressed for each segment can be interpreted as a formation identifier by clustering the average role adjacency matrix, labeling a formation identifier for each cluster, or using an artificial neural network trained with data in which position information is labeled with a formation identifier. You will be able to.

In the RoleCPD step, a change point within the formation period can be detected. At this time, the change point can be detected from a sequence of role permutation within the formation period, and various change point detection algorithms described above can be used. For example, similar to the FormCPD step, a role change point can be obtained by applying discrete g-segmentation to a sequence of permutations with a pairwise Hamming distance. Role change points can divide each formation period into several role periods. Here, the role period may refer to a period during which the role assigned to the player is maintained.

Also in the RoleCPD phase, each player is assigned one dominant role per role period to find out their actual position (e.g. "Center Forward" or "Right Midfielder") for each role period. can

Also, in the RoleCPD step, the positional transposition of the players can be calculated. Here, the position conversion may be derived from the role substitution sequence in each formation period.

On the other hand, some high-level tactical analysis can be performed using SportsCPD.

According to an embodiment of the present disclosure, analysis of a switch pattern may be performed using SportsCPD. Here, the switch may mean exchanging roles between players. Analysis of switch patterns can be performed by considering player-role assignments. In more detail and without limitation, the role exchange between players, that is, a position change, may include only a temporary role exchange between players while the same formation is maintained, except for a position change that occurs when the team formation is changed. there is. In this case, analysis of the switch pattern may be performed in consideration of player-role assignment within the formation period.

Specifically, the analysis of the switch pattern may be performed as follows.

For example, first, a main player-role assignment and a sub-player-role assignment may be obtained within a formation period. Since formations are fixed within a formation period, only one primary player-role assignment can be determined. For example, among the player-role assignments corresponding to the formation period, the most frequent player-role assignment may be determined. Next, the secondary player-role assignments can be determined. For example, among the player-role assignments corresponding to the formation period, whether the number of players excluding the main player-role assignment is within a predetermined number in order of highest frequency, whether the number is greater than or equal to the predetermined frequency, and the ratio of the occurrence frequency during the formation period Sub player-role assignment can be determined depending on whether or not the number is above a certain level and the conditions in which they are combined. Once the primary player-role assignment and the secondary player-role assignment are determined, switch patterns can be analyzed based on them. For example, based on the difference between the main player-role assignment and the secondary player-role assignment, that is, the role of each player changed between the two assignments, the player participating in the switch or the role participating in the switch may be determined. As another example, the number of trials or the tendency of trials of a corresponding switch may be determined based on the occurrence frequency of secondary player-role assignment. Meanwhile, in the case of a target session having a plurality of formation periods, a switch pattern for the entire target session may be analyzed by performing the above-described process for each formation period.

As another example, a switch pattern may be analyzed based on a divided role period within a formation period. First, representative player-role assignments can be extracted for each role period. The representative player-role assignment for each role period may be the most frequent player-role assignment within the role period. If the representative player-role assignment is extracted, it is possible to determine the player and/or role participating in the switch based on the main player-role assignment of the formation period and the representative player-role assignment by role period, or to determine the number of attempts or the tendency of attempts. can Alternatively, after selecting the most frequent assignment among representative player-role assignments, the above-described information may be obtained based on other representative player-role assignments. Alternatively, the above-described information may be obtained based on representative player-role assignments adjacent to each other. Meanwhile, in the case of a target session having a plurality of formation periods, a switch pattern for the entire target session may be analyzed by performing the above-described process for each formation period.

In addition, according to an embodiment of the present disclosure, it is possible to perform detection of an irregular situation such as a set-piece using SportsCPD. For example, based on the main player-role assignment for each scene based on the main player-role assignment during the formation period, it may be determined whether the corresponding scene is an irregular situation.

SportsCPD according to an embodiment of the present disclosure may have the following several advantages. First, it is possible to detect the team's basic tactics and their changes in an unsupervised way by applying an advanced changepoint detection method to sports tracking data. Second, the formation structure and role switch for an arbitrary time interval can be effectively expressed as an average role adjacency matrix and a sequence of permutations, respectively. Third, it can be used as a practical application that can be easily understood and used by workers in the field, such as switch pattern search and automated set-piece detection.

3. 관련 기술3. Related technology

3.1 비모수적 변화점 검출3.1 Non-parametric change point detection

Change-point detection (CPD) is a technique for finding a point at which governing parameters in a series of processes change. CPD can be classified into a parametric method and a nonparametric method according to whether parameters are estimated in detail or points of change are directly found.

In particular, the graph-based non-parametric method is considered to be more flexible in practical application because it uses only similarity information and can be applied to high-dimensional data or even non-Euclidean data. An example of CPD is g-segmentation, which defines a graph-based scan statistic. Using this, it is possible to draw a similarity graph such as a minimum spanning tree for the observations and check the number of segments connected to the observations before and after each time t. In this case, when the number of line segments is significantly small, t may be regarded as a potential change point. In addition, techniques for correcting g-segmentation errors that occur when there is a change in both the average and the deviation or when the change point is not in the center of the sequence can be considered.

However, the above-described methods have a limitation in that they can be applied only to continuous data. For example, a similarity graph for data having repeated observations may not be uniquely determined by the graph-based method described above. To address this and obtain a unique scan statistic, a combination of multiple optimal similarity graphs or a discrete g-segment using the average of the resulting statistic can be applied. It has been experimentally demonstrated that discrete g-segmentation works well for discrete data, such as network data.

3.2 팀 스포츠에서 포메이션과 역할의 추정3.2 Estimation of formations and roles in team sports

The following several studies dealt with spatio-temporal tracking data to estimate team formation and find positions, that is, players' roles in sports games. Balkowski's thesis "Large-scale analysis of soccer matches using spatiotemporal tracking data (2014)" proposes a role representation algorithm that dynamically assigns a unique role to each player for each scene. This algorithm is similar to the K-means algorithm, except that the Hungarian algorithm is used in the E-stage to satisfy the constraint that no two players play the same role at the same time. Narizukawa Yamazaki's paper "Characterization of the formation structure in team sports (2017)" presents the provisional formation as an undirected graph using Delaunay triangulation and clustering the graph for a single game to characterize the formation structure. do. Narizuka and Yamazaki's follow-up paper "Clustering algorithm for formations in football games (2019)" combines role representation techniques to cluster formations across multiple matches.

The approaches described above cannot distinguish between a fixed role change by tactical instruction and a temporary role swap, which is mostly unintended from a macro perspective. In an actual team sports game environment, changes according to the coach's instruction are generally at most several times throughout the game or not, and role changes are not instructed every short time interval (eg, every second). Applying the change point detection algorithm according to an embodiment of the present disclosure to the tracking data can detect “actual” role changes between players.

4. 문제 정의4. Define the problem

In team sports, coaches often change team formations a few times during a game, or instruct players to change roles without changing formations. Therefore, their decision-making process can be traced by following two steps. First, find the point of change in formation, and then find the point of change in the next role. In this section, each step will be treated as a separate change point detection problem with reference to FIG. 1 .

1 is an example of a timeline of formation and role changes according to the results of SportsCPD according to an embodiment of the present disclosure. In Figure 1, the upper figure shows the team formation for each formation period. Each colored dot represents the scene-specific coordinates of a specific role normalized by the team's average position. White circles are the average positions of individual roles, and the thickness of the line connecting the circles corresponds to the corresponding element of the average role adjacency matrix. The timeline at the bottom shows the roles assigned to players per role period.

4.1 포메이션 변화점 검출 4.1 Formation Change Point Detection

The team formation can be considered using the probability distribution of the role phase position relationship. The goal of FormCPD is to find the point of change in the distribution for a set of observations in the form of bow trajectories. Therefore, it is possible to first find an effective representation of the role-phase-location relationship from raw observations, and then perform CPD using the sequence of feature factors. After finding the feature factor, the change point detection can be performed using the sequence of feature factors.

Expressing this as an equation, when the sequence of feature factors {A(t)} _t∈T representing the role phase position relationship is known, several distributions F ₁ , . . . _, T ₁ < _. <T _m can be obtained. Here, each interval T _i may be referred to as a formation period in which a team has a unique formation represented by a distribution F _i within that period.

4.2 역할 변화점 검출4.2 Detecting Role Change Points

Although Balkowski's thesis described above proposes a role expression that allocates roles for each scene to field players excluding the goalkeeper for a soccer game, long-term tactical changes and temporary role exchange cannot be distinguished. Given that players typically play fixed roles directed by coaches over a period of time, the goal of RoleCPD is to find points of change where long-term tactical changes occur between before and after. That is, the formation period T _i is several time intervals T _i,1 < . . . in which players have fixed roles. <T _i,ni , which may be referred to as a role period. Then, mutual change between roles for each scene according to the result of role expression within the role period may be regarded as a temporary swap.

Expressing this as a formula, for a set P of players at game time T, a set of roles satisfying the following formula χ={X ₁ , . . . , X _N } and the player's temporary role map (P-TR map: Player-to-Temporary-Role map) {β _t :P→χ} _t∈T .

β _t (p)≠β _t (q) ∀p≠q∈P, t∈T

Here, the temporary role map {β _t } _t∈T is a player's designated role map (P-IR map: Player-to-Instructed-Role map) {α _t :χ→χ} symmetric group for _t∈T and χ ( It can be expressed as a combination of temporary role permutation {σ _t :χ → χ} _t∈T with σ _t ∈S(χ) for the symmetric group S(χ). Specifically, β _t = σ _to α _t In other words, for a P-TR map {β _t } _t∈T for a formation period T _i , T _i _,1 < . <T _i,ni and the P-IR map {α _t } _t∈T can be found.

The first is periodwise consistency, in which the indicated role for every player is fixed in each role period T _i,j . For example, for some X _p ^{(i, j)} ∈ χ, the following equation can be satisfied.

α _t (p)=X _p ^(i,j) ∀t∈T _i,j

The second is uniqueness: no two players are assigned the same role within a single role period. For example, the following formula can be satisfied.

X _p ^(i,j) ≠X _q ^(i,j) ∀p≠q∈P

The third is the existence of role change. A change in the duration of a role can mean a change in instruction. For example, the following formula can be satisfied.

X _p ^(i,j) ≠X _p ^(i,j+1)

The last condition can only be valid for changes within a single formation period. That is, there may be identical player-role assignments in adjacent role periods of distinct formation periods. An example of this is shown in the first two role periods in FIG. 1 .

5. 방법론5. Methodology

This section describes in turn the detailed process of dividing a match session into multiple formation periods (FormCPD) and each formation period into multiple role periods (RoleCPD). It should be noted in advance that the methodology described below is a non-limiting example.

5.1 역할 인접성 행렬에 기반한 포메이션 변화점 검출5.1 Detection of formation change points based on role adjacency matrix

First, for each scene, the role phase position relationship can be expressed as a role adjacency matrix calculated by Delaunay triangulation. Then, a formation change point can be found using the pairwise distance between the role adjacency matrices through discrete g-splitting, which is a non-parametric change point detection algorithm. The formations in each segment can be represented by an averaged role adjacency matrix, which is then computed by dividing the entire data set with labels used in the field, such as "3-4-3" or "4-4-2", assigned to individual formation periods. can be clustered according to

5.1.1 역할 인접성 행렬의 시퀀스 생성5.1.1 Creating a sequence of role adjacency matrices

First, roles may be assigned to players for each scene according to a role expression method. Here, the role expression may be exemplarily performed by referring to the concept presented in the above-mentioned Balkowski paper. An adjacency matrix can then be obtained from the role representation. Here, as an example, the adjacency matrix may be {A(t)} _t∈T ⊂ R ^N×N obtained between role labels according to Delaunay triangulation. At this time, the component a _kl (t) is the role group χ={X ₁ , . . . ,X _N }, it may be 1 if X _k and X _l are adjacent at t, and may be 0 otherwise. Figure 2 shows an example of a Delaunay graph for role positions.

2 and 3 are graphs of role adjacency according to an embodiment of the present disclosure. FIG. 2 is a temporary Delaunay graph for player trajectories in an arbitrary scene, and FIG. 3 is a weighted graph drawn from average role positions and an average role adjacency matrix for a formation period. The numbers in each circle indicate role labels 1 to 10, and the numbers under the circle in FIG. 2 are the corresponding player's uniform number.

The application of Delaunay triangulation to player tracking data was first proposed by Narizuka and Yamazaki. However, they used the back number as the index of the adjacency matrix, and applied the role representation only if the back numbers across multiple games matched a single set of indices. According to this, position exchange patterns can be explored well, but as a result player adjacency matrices will be greatly affected by irregular situations such as temporary switches or set-pieces where the original team formation collapses. can Therefore, a more robust representation of the formation structure can be obtained by using role labels as an index of the adjacency matrix instead of these player labels.

5.1.2 행렬 시퀀스에 대한 이산 g-분할의 적용5.1.2 Application of discrete g-partitioning to matrix sequences

In order to find formation change points, CPD can be applied to sequences of role adjacency matrices obtained from each session (eg, first half or second half) of a game, respectively. Among various CPD methods, discrete g-segmentation has the distinction that it can be applied to high-dimensional or non-Euclidean data with repeated observations. Given that the sequences of the embodiments of the present disclosure are frequently repeated high-dimensional observations (e.g., a 10×10 binary matrix computed from 10 field players for soccer), there is high confidence in applying the discrete g-segment. can be expected

First, g-segmentation may be exemplarily performed as follows. First, we can construct a similarity graph G, such as a minimum spanning tree, for given observations. Then, for each time stamp, the lines are a type-0 edge connecting observations before and after t, and a type-1 edge connecting observations before t. It can be classified into three types: , and type-2 edges connecting observations after t. Accordingly, R ₀ (t), R ₁ (t), and R ₂ (t), which are the numbers of corresponding types of line segments, can be obtained. From this, the generalized line-count scan statistic can be defined as follows when Σ(t) is the covariance matrix of (R ₁ (t), R ₂ (t)) ^T under the null distribution obtained by mixing the time index .

S(t)=(R ₁ (t)- Ε [R ₁ (t)], R ₂ (t)- Ε [R ₂ (t)]) ^T Σ ^-1 (t)(R ₁ (t)- Ε [R ₁ (t)], R ₂ (t) - Ε [R ₂ (t)]

Since similar observations are more likely to be connected by line segments than distinct observations, a large S(t) can mean small intra-group distances and large inter-group distances (where "groups" is a time interval separated by t). indicated). In this way, according to the algorithm, when S(τ) exceeds a predefined threshold, τ=arg max _t S(t) may be determined as a change point.

However, the similarity graph may not be uniquely defined in g-segmentation, and thus g-segmentation may not be suitable for data processing with repeated observations. Therefore, in the discrete g-partition, instead of (R ₁ (t), R ₂ (t)), averaging statics (R _1,(a) (t), R _2,(a) (t) or combination statistics ( union statics (R _1,(u) (t), R _2,(u) (t)) can be used to overcome the drawback of the g-partition. and a similarity graph G ₀ created by the number of each subsequent observation.

When obtaining a sequence of matrices by applying discrete g-partition, the following L1,1 matrix norm (hereinafter referred to as 'Manhattan distance') can be used to measure the distance between role adjacency matrices.

d _M (A(t), A(t'))=||A(t)-A(t')|| _1,1 =Σ _k=1toN Σ _l=1toN |A _kl (t)-A _kl (t')|

Also, if the temporary role assigned to the player in each scene t is different from their most frequent role throughout the session, it can be determined that the player is switching roles in each scene t. The switch rate at t can be calculated from the number of players switching, and scenes with a high switch rate (e.g., 0.7 or higher) have ratios such as set-pieces in which players do not stay in formation at all. may be excluded under valid circumstances. In summary, the discrete g-segmentation can return the predicted change points among the sequence of remaining valid adjacency matrices using pairwise Manhattan distances.

Once the change points are found for a given time period, the algorithm can determine their significance according to the three conditions individually or in combination. As a specific example, (1) the p-value of the scan statistic is less than 0.01, (2) two segments last more than a minimum time (eg, 5 minutes), and (3) the calculated Only when the average role adjacency matrices are far enough from each other (i.e., the Manhattan distance is large) can the estimated change point τ be considered significant. The critical distance for (3) can be empirically set to 7.0.

Since there may be more than one formation changepoint during a session, a recursive framework can be built to find multiple changepoints. First, if there is a significant change point within a given period, the CPD algorithm can be applied to sequences before and after the change point, respectively. Each branch CPD (branch Change-Point Detection) may end when there are no significant change points in the segment of interest. Consequently, for a given session T, several formation periods T ₁ <... <T _m .

5.1.3 평균 역할 인접성 행렬에 기초한 포메이션 군집화5.1.3 Formation clustering based on the average role adjacency matrix

Based on the above, the formation in the formation period T _i is the average role position V(T _i )=|T' _i | ^-1 Σ _t∈T'i V(t), and as an edge matrix, the adjacency matrix A(T _i )=|T' _i | When ^-1 Σ _t∈T'i A(t), it can be calculated as a weight graph F(T _i )=(V(T _i ), A(T _i )). Here, T′ _i is a set of effective time points with a low switch rate among T _i , and V(t)=(v ₁ (t), …, v _N (t)) ^T ∈R ^N×2 is each scene t is a two-dimensional position of N roles normalized such that the average is '0' (eg, Σ _{k = 1toN} v _k (t) = (0,0)) in . 3 is an example of a formation graph.

Here, the distance between arbitrary formation pairs may be defined by arranging the roles and calculating the Manhattan distance between the aforementioned role adjacency matrices. Specifically, let's assume that any pair of formations F=(V, A) and F'=(V', A') in the formation period. Here, F and F' do not necessarily belong to the same game, and they may be calculated in different games. Next, the rows and columns of A can be rearranged into the rows and columns of A'. For example, a Hungarian algorithm can be used for rearrangement. In this case, the pairwise Euclidean distance between V and V' ) can be used. In other words, the assignment cost c(V, V'; Q)=Σ _k=1toN ||(QV) _k -v' _k || An optimal permutation matrix Q that minimizes ₂ can be calculated. For example, the distance between formations F and F' is d(F, F')=d _M (QAQ ^T _, A') QAQ ^T and, as in the case ofA' can be used.

Depending on these distances, we can perform clustering of formations to determine whether the formations are identical or different using the entire data set. If agglomerative clustering is applied to the pairwise formation distance, the formation graph can be grouped into a plurality of clusters. Labeled with terms familiar to those working in the field.

4 relates to an example formation group according to an embodiment of the present disclosure. Figure 4 shows the average role position of each formation group, and terms used in the field of soccer are labeled for this, and the numbers in parentheses are the proportion of the total playing time. The result of FIG. 4 is calculated using positional data obtained from GPS data as raw data, that is, player tracking data. Specifically, the results of FIG. 4 show that for 809 sessions of 2.33 million seconds, SportsCPD is applied to divide them into 866 formation periods and 2,158 role periods, and then cluster the 866 formation periods into 32, which It was merged again into a group of 7 formations.

While the method proposed by Balkowski uses Earth Mover's Distance (EMD) between role locations, the clustering according to the above-described embodiment of the present disclosure may use the average distance between role adjacency matrices. In addition, in the prior art, only formations for half of the game with an almost constant length of time were dealt with, but in the embodiment of the present description, the formation period is shorter than this or the length varies widely, which is likely to cause distortion of the formation graph. You can handle the formation. This is because the average role adjacency matrix based on adjacency relationships rather than absolute position is more robust against such distortions.

5.2 역할 치환열에 기반한 역할 변화점 검출5.2 Role Change Point Detection Based on Role Substitution Sequence

RoleCPD can be performed in a similar way to FormCPD, except that a role permutation sequence is used instead of a role adjacency matrix. For each formation period, a discrete g-segment can be applied recursively to find role changes. Next, the original role indicated to each player for each role period can be found, and the player-role assignment for each scene can be expressed as a substitution sequence of the designated role.

5.2.1 역할 치환열 생성5.2.1 Generation of role substitution sequences

In the above-described role expression, each player p may be assigned an initial role X _p according to a standard number such as a back number, for example. Using this, the players at time t p ₁ , … , the temporary role assignment for p _N is the substitution sequence for the initial role assignment (X _P1 , … , X _PN ) π _t =(π _t (X _P1 ), … , π _t (X _PN ))∈S( χ) can be expressed as Expressing this again as a formula, it can be as follows.

β _t (p)=π _t (X _P )

5.2.2 치환열의 시퀀스에 대한 이산 g-분할의 적용5.2.2 Application of discrete g-partitioning to sequences of substitutions

Similar to the application to the role adjacency matrix, g-segmentation can be applied to the sequence of role permutation sequences {π _t } _t∈Ti for each formation period T _i . According to this, the Hamming distance d _H (π _t, π _t' )=|{X:π _t (X)≠π _t' (X), X∈χ}| is used as the distance between the role substitution sequences A predicted change point can be determined among sequences of effective substitution sequences (for example, substitution sequences when the switch rate is 0.7 or less).

Similar to FormCPD, a test for the significance of the change point can be performed, and at this time, the last condition applied to FormCPD can be excluded. Since the goal of RoleCPD is to find when the main role assignment is changed, the change point τ can be judged to be significant only when the most frequent substitution sequence before and after τ is different for the change point τ.

Finally, by applying a recursive CPD to the sequence in each formation period T _i , the partition T _i,1 < . . . of T _i . <T _i,ni can be obtained.

5.2.3 역할 기간 별 선수의 지정 역할 도출5.2.3 Derivation of designated roles for players by role period

For each role period T _{i,j ,} the most frequent permutation sequence for the player can be set as the designated role, and all temporary roles obtained from the role expression can be expressed as the permuted sequence of the designated role. Expressed as a formula, the P-IR map {α _t :P→χ} _t∈Ti,j is exemplarily, π _(i,j) ∈S(χ) as {π _t } _t∈Ti,j As a frequent substitution sequence, it can be expressed as α _t (p)=π _i,j (X _p ), which is a fixed value at t∈T _i,j . In t∈T _i,j, the role substitution sequence σ _t (X _p )=β _to α ^-1 can be obtained as follows.

σ _t (X _p )=β _t( α _t ^-1 (X _p ))=π _t (π _(i,j) ^-1 (X _p ))

Accordingly, the P-IR map may satisfy the three conditions described above. Specifically, α _t (p) is constant at t∈T _i,j and it is distinct across p∈P because it is a substitution sequence with a distinct initial role and thus has temporal consistency and uniqueness. In addition, the existence of a role change between adjacent role terms can be secured due to the significance test.

6. 실험6. Experiment

The accuracy of SportsCPD can be evaluated by comparing the result according to the embodiment of the present disclosure with the measured value recorded by an expert in the field. In addition, according to an embodiment of the present disclosure, an interpretation technique for identifying play patterns of a team in terms of a role switch may be provided. In addition, the switch rate according to the embodiment of the present disclosure can be used as a simple and highly reliable index for set-piece detection.

6.1 모델 평가6.1 Model evaluation

The performance of SportsCPD can be measured in two aspects. One is the accuracy of prediction formation by FormCPD, and the other is the robustness of role change point detection according to RoleCPD. Regarding the former, it is possible to calculate the ratio of correctly detected segments per minute to the total number of segments (eg, total play minutes). Regarding the latter, it is measured by how accurately RoleCPD can detect the change point, given that change point detection is about the classification of each time point between the change point and the non-change-point. You can try it.

For both evaluations, ground truth labels recorded by video analysts or subject matter experts working as coaches for professional or semi-professional sports teams may be used. Since it is very exhausting to record formations and roles by minutes, the recording of one game by an expert was used for the evaluation according to the embodiment of the present disclosure.

For formation prediction, comparison was made with labels for each minute segment instead of each formation period, because the method according to the embodiment of the present disclosure and the change point detected by a human expert are slightly different. Out of a total of 2,680 minutes of playing time, the record for 2,047 minutes and the result of FormCPD were consistent, showing a degree of agreement of 76.4%. 5 shows detailed results as a chaos matrix.

5 is a diagram related to formation prediction according to an embodiment of the present disclosure. Figure 5 expresses the accuracy of formation prediction as a confusion matrix, which compares the predicted formation in minutes with the actual recorded formation.

On the other hand, RoleCPD detected 96 change points in 28 games, 86 change points were located within 5 minutes of the recorded change points, and there were a total of 7 missing change points (false negatives). Therefore, the recall points and F1 scores for accuracy of RoleCPD are 0.8958 (86/96), 0.9247 (83/93), and 0.9101 (2·(precision×recall)/(precision+recall)), respectively.

6.2 스위치 패턴 검색6.2 Switch Pattern Search

The role substitution sequence obtained as described above may indicate a temporary role switch between players. In the case where all players maintain their assigned original roles at time t, the role permutation sequence σ _t can be an identity permutation. In other words, the non-identity role substitution sequence may mean that a switch play occurred at that time. Therefore, by analyzing the non-identity role substitution sequence, the play pattern of the team can be identified.

6 is a graph of role substitution sequences for each role period according to an embodiment of the present disclosure. In FIG. 6, the period of the top 5 role substitution sequences with the highest frequency for each role period is expressed in seconds. 7 is a diagram related to an example of tactic analysis using a role substitution sequence according to an embodiment of the present disclosure.

Referring to the results of FIGS. 6 and 7 for the game shown in FIG. 1 , it is clear that a team's tactical pattern can be identified from the role substitution sequence for a specific game. Figure 6 shows the five most frequent role substitution sequences for each role period, and the following analysis was performed from them. In the following analysis, the formation period and the role period are referred to as FP (Formation Period) and RP (Role Period), respectively, and a player assigned a specific role k will be denoted as Rk.

However, players actively perform role switches in RP1 (in this section, the sum of the time of the top 3 role permutation sequence is 140 seconds out of 18 minutes), whereas in RP3 (the top 3 role with high frequency in this section) The total time of the heat of substitution was only 35 seconds out of the total 13 minutes), almost no switch was made.

Second, the role substitution sequence (3 5) occurred frequently in RP1, which is a "false-9 play" in which the center forward shown in Figure 7 falls into the back of the midfielder to provide space for fellow players. " means However, in RP2, the team formation changed to 4-3-3 and the play was reduced.

Third, unlike other RPs, in RP1, the fullbacks (R1, R8) took on the attacking role along the side and actively overlapped with the wingers (R9, R10). Here, it is revealed that the overlap is different on both sides. On the left side, R6 covers the position of R8, resulting in a 3 cycle of (6 8 9), while on the right side, R4 does not cover the position of R1. A two-factor cycle of (1 10) occurred.

Fourth, in RP2, left winger R9 (6 9) tried to cut-in play towards the penalty box through the switch.

Fifth, in RP4, after adding the center forward R5, the team can switch between the center forwards (R3 R5) or between the central midfielders (R4 R6) except for the usual switches (i.e. switches within the same position). kept the formation constant.

6.3 세트-피스 검출6.3 Set-piece detection

In a sports game, a set-piece may refer to a situation in which a game is stopped and restarted. In particular, in soccer, in set-piece situations, you can send the ball into the scoring area, and set-pieces such as free kicks and corner kicks are considered good opportunities to score points. The review is meticulous, and special tactical training for set play is also conducted.

SportsCPD's output automatically detects set-piece situations using statistical switch rates, allowing sports teams to easily extract and manipulate set-play data. The term switch rate has been defined as the ratio of players switching in the above, but it can also be defined as a value obtained by dividing the distance to the identity substitution sequence (eg, Hamming distance) by the number of roles. Because players are completely mixed up in set-piece situations, switch rates in these situations can be close to 1.0. Thus, using the switch rate provides a simple, completely unsupervised, yet highly accurate set-piece detection model. For example, a set-piece may be detected by selecting a situation where the switch rate is greater than or equal to a threshold value (eg, 0.9). 8 shows a strong cross-correlation between switch rate and set-piece.

8 is a graph illustrating a relationship between a switch rate and set-piece occurrence according to an embodiment of the present disclosure. FIG. 8 is a time-sequential diagram of switch rates and set-piece occurrences during the first half of the game shown in FIG. 1 . In FIG. 8, "C" and "F" indicate a corner kick and a free kick that can be kicked into the opponent's penalty box, respectively. The white square means the case where the team that calculated the switch rate in the situation is the attacking team, and the black square means the case the defending team.

7. 구현7. Implementation

Hereinafter, a device, system, and method for implementing SportsCPD according to an embodiment of the present disclosure described above will be described.

9 is a block diagram of a system for providing tactical analysis of team sports according to an embodiment of the present disclosure.

Referring to FIG. 9 , a system 1000 may include a sensor device 1200 , an analysis device 1400 and a user device 1600 . Hereinafter, the components of this system will be described in more detail.

The sensor device 1200 may be provided in the form of an attachable device. Here, the term "attachable device" may refer to a device directly or indirectly attached to the player. For example, the attachable device may be a pocket device inserted into a player's clothes or a wearable device in the form of a band wrapped around a sports player's body part, such as a wrist or ankle. Since the sensor device 1200 in the form of an attachable device is attached to each player, there may be a plurality of them in team sports.

For example, the sensor device 1200 may perform global localization on a player. In this case, the player tracking module 1220 may be a global navigation satellite system (GNSS) module including a global positioning system (GPS) module. For example, the GPS module performs global positioning of the player by calculating the GPS global position (eg, latitude, longitude) through a triangulation technique by a GPS processor using GPS signals received from navigation satellites through a GPS antenna. can do. The calculated global location may be transmitted to the analysis device 1400 through the communication module 1240 . Accordingly, the analysis device 1400 may obtain a player tracking data set. In addition, during this process, the analysis device 1400 may convert the global position received from the sensor device 1200 into a field coordinate system.

As another example, the sensor device 1200 may perform local positioning on a player. In this case, the sensor device 1200 may operate as a tag node of a Local Positioning Network (LPS) and transmit/receive an LPS signal with an anchor node fixedly installed in a positioning area. In the case of performing global positioning, the sensor device 1200 can independently perform positioning, whereas in the case of performing local positioning, the sensor device 1200 simply functions as a transmitter/receiver for the LPS signal of the local positioning network. And the positioning result may be calculated in an external device.

For example, the sensor device 1200 may transmit or receive an LPS signal to an anchor node, and positioning of a player may be performed by an external device using a result of transmitting and receiving an LPS signal between nodes in a network. At this time, the analysis device 1400 receives the result of transmitting and receiving the LPS signal and directly performs positioning to generate player tracking data, or may receive the positioning result from a separate external device that performs positioning using the result of transmitting and receiving the LPS signal. .

Meanwhile, unlike shown in FIG. 9 , in the tactical analysis system 1000, the sensor device 1200 may be replaced with a video analysis platform. The video analysis platform may be composed of a camera that captures an image and a video analysis device that calculates the position of an object (eg, a player or a ball) in an image through video analysis using top-view conversion and deep learning algorithms. . At this time, the analysis device 1400 may directly receive an image from a camera and perform image analysis to generate a player tracking data set, or may receive a positioning result from a separate external device that performs image analysis.

The analysis device 1400 may perform various operations described above in relation to SportsCPD according to an embodiment of the present disclosure. Basically, unless otherwise noted, various operations described above in relation to SportsCPD can be interpreted as being performed by the analysis device 1400, and among them, it is revealed that they can be interpreted as being performed by the controller 1440. . For example, the analysis device 1400 receives information directly or indirectly reflecting the player's position from the sensor device 1200, generates a player tracking data set therefrom, and uses it to generate the above-described SportsCPD and related data sets. It can perform various applications (detection of irregular situations, etc.). This analysis device 1400 may be provided in the form of a personal computer or a local or remote server, and may not necessarily be physically provided as a single entity.

The analysis device 1400 may include a communication module 1420 , a controller 1440 , and a memory 1460 .

The communication module 1420 may transmit/receive data between the analysis device 1400 and an external device (eg, the sensor device 1200 or the user device 1600). For example, the analysis device 1400 may collect data from an attachable device through the communication module 1420, receive an image from a camera, or transmit various types of information to the user device 1600 through the web.

The controller 1440 may control overall operations of the analysis device 1400 . The controller 1440 may be implemented as a hardware configuration, a software configuration, or a combination thereof. From a hardware point of view, the controller 1440 may be provided in various forms capable of performing calculations or data processing, including an electronic circuit, an integrated circuit (IC), a microchip, and a processor. In addition, since the physical configuration of the controller 1440 is not necessarily limited to a single physical entity, the controller 1440 is a single processor that integrates all processes of the analysis device 1400 or It may be provided as a plurality of processors each performing different functions, or may be provided in a form combined with some of the other components of the analysis device 1400.

The memory 1460 may store various data related to the operation of the analyzer 1400 . The memory 1460 may be provided with various volatile and non-volatile memories.

The user device 1600 may function as a user interface that provides various data or information collected or calculated by the system 1000 to the user or receives user input from the user. For example, the user device 1600 receives a tactical analysis result from the analysis device 1400 through the communication module 1620 under the control of the controller 1640 and provides it to the user as visual information through the display 1660. can do. The user device 1600 may be a smart device such as a smart phone or tablet, a personal computer such as a laptop or desktop, or an electronic device similar thereto.

Hereinafter, various examples of the aforementioned analysis method according to an embodiment of the present disclosure will be described with reference to FIGS. 10 to 18 . 10-18 are flow charts of examples of methods for providing tactical analysis of team sports according to embodiments of the present disclosure.

The methods described below may be used alone or in combination with each other, and since not all of the steps of the methods are essential, the methods may be performed including some as well as all of the steps. In addition, since the order in which each step in the method is mentioned is only for convenience of explanation, the steps do not necessarily have to be performed in the order in which they are described. In addition, hereinafter, each method will be described based on what is performed by the above-described system and device, but since this is also only for convenience of explanation, each method is a tactical analysis system 1000 and a tactical analysis device 1400 It is not something that can only be done by

Referring to FIG. 10 , an example of a method for providing tactical information for team sports according to an embodiment of the present disclosure includes obtaining a player tracking data set (S1100) and obtaining a player-role assignment (S1200). ), obtaining a role arrangement (S1300), detecting a change point from a sequence of role arrangements (S1400), and dividing a target session into formation periods (S1500). Hereinafter, each step of this example will be described.

The analysis device 1400 may obtain a player tracking data set (S1100).

Specifically, the controller 1440 may receive a player tracking data set from the sensor device 1200 through the communication module 1420 . When the sensor device 1200 performs global positioning, the analysis device 1400 may receive player position information expressed as a global position during the target session and convert it into a stadium coordinate system. When the sensor device 1200 performs regional positioning, the analysis device 1400 may receive a transmission/reception result of the LPS signal from the local positioning network, calculate the position of the player using the LPS signal, and obtain a tracking data set. Alternatively, the analysis device 1400 may obtain a player tracking data set by receiving an image captured by a camera and calculating a player position therefrom, or by receiving a player position calculated by an external device from a camera image.

On the other hand, the player position corrects the position of each player using the positions of all or part of the players participating in the target session in the scene (for example, the same team, field players excluding the goalkeeper, etc., or a combination thereof) can do. For example, each player's corrected position may be corrected as an average position in a corresponding scene of field players of the same team.

Next, the analysis device 1400 may use the player tracking data set to obtain player-role assignments (S1200). For example, the controller 1440 initially assigns a single random number to the players throughout the session. Allocate roles, obtain position distribution of roles based on this, perform player-role assignment for each scene in consideration of position distribution of roles, recalculate position distribution of roles in consideration of player-role assignment for each scene again, The player-role assignment can be obtained through a process of re-executing the player-role assignment for each scene based on the recalculated position distribution. In this process, the controller 1440 may exclude scenes corresponding to non-regular situations when calculating the location distribution of roles. Whether the scenes correspond to non-regular situations will be described in detail in an embodiment to be described later.

Next, the analysis device 1400 may obtain a role arrangement using the player-role assignment (S1300). For example, the controller 1440 may obtain the role arrangement using the location distribution of the roles, which has been described in detail above.

When the role arrangement is acquired, the analysis device 1400 may detect a change point from the sequence of the role arrangement (S1400), and may divide the target session into formation periods using this (S1500). Specifically, the controller 1440 may apply a change point detection algorithm to a sequence of role assignments to detect one change point, and based on this, divide the target session into pre- and post-periods, and divide the divided periods again. The formation period can be obtained by performing re-division by applying the changing point detection algorithm. Furthermore, the controller 1440 may determine whether the change point detected by the change point detection algorithm is valid based on several conditions, which have been described in detail above.

Referring to FIG. 11 , an example of a method for providing tactical information for team sports according to an embodiment of the present disclosure includes obtaining a player tracking data set (S2100) and obtaining a player-role assignment (S2200). ), detecting a change point from the sequence of player-role assignment (S2300), and dividing the target session into role periods (S2400). Hereinafter, each step of this example will be described.

First, the analysis device 1400 may obtain a player tracking data set (S2100). A process of acquiring player tracking data may be performed similarly to step S1100. However, at this time, the player tracking data set may be related to the entire target session, or may be a single formation period. If the entire session is divided into a plurality of formation periods, the method of this example may be repeatedly performed for each formation period.

Next, the analysis device 1400 obtains the player-role assignment (S2200), detects a change point from the sequence (S2300), and divides the target session into role periods (S2400). Specifically, the controller 1440 may obtain the player-role assignment for each scene similarly to that described in step S1200, and may determine the changing point by applying a changing point detection algorithm to the sequence. Even at this time, the controller 1440 may determine the validity of the change point through several conditions (detailed descriptions thereof have been described above), and may divide the role period based on the valid change point.

Referring to FIG. 12 , an example of a method for providing tactical information for team sports according to an embodiment of the present disclosure is, in the method according to FIG. 11 , a main player for a role period-determining a role assignment (S3500 ), determining whether the player-role assignment is a non-regular player-role assignment (S3600), calculating a switch rate for the non-regular player-role assignment (S3700), and detecting a non-regular situation in the target session based on the switch rate A step including step S3800 may be further included. Hereinafter, each step of this example will be described.

In this example, the analysis device 1400 may determine the main player-role assignment for each session divided into role periods (S3500). Specifically, the controller 1440 may determine the most frequent player-role assignment during the role period as the main player-role assignment.

When the main player-role assignment is determined, the analysis device 1400 determines whether the player-role assignment for each scene is a non-regular player-role assignment based on this (S3600), and calculates a switch rate for the non-regular player-role assignment. (S3700) Specifically, the controller 1440 may determine whether the corresponding scene corresponds to an irregular situation based on the difference in player-role assignment for each scene with respect to the main player-role assignment. For example, when a player-role assignment is expressed as a role permutation sequence, when the distance between the two (eg, Hamming distance, etc.) is greater than or equal to a threshold value, or when the two are different, the player-role assignment is converted to a non-regular player-role. It can be judged by assignment. The controller 1440 may calculate a switch rate for the non-regular player-role assignment, and the switch rate is a value reflecting the difference between the non-regular player-role assignment and the main player-role assignment. A detailed description of this is described above. there is a bar

If the switch rate for each scene is calculated, the analysis device 1400 may detect an irregular situation in the target session based on it (S3800). For example, the controller 1440 may determine a scene having a switch rate equal to or greater than a threshold value as an irregular situation.

Referring to FIG. 13 , an example of a method for providing tactical information for team sports according to an embodiment of the present disclosure includes, in the method according to FIG. 10 , acquiring location information for each role related to a formation period (S4600). , and identifying a formation based on location information for each role (S4700) may be further included. Hereinafter, each step of this example will be described.

In this example, when the formation period in which the analysis device 1400 divides the target session is obtained, position information for each role may be obtained (S4600). Here, the acquisition of the formation period may be performed, for example, by the analysis device 1400 according to the method described with reference to FIG. 10 , and the analysis device 1400 based on this obtains location information for each role over the entire session. can be obtained For example, the controller 1440 may refer to the player tracking data set and the player-role assignment for each scene, determine the position of the role for each scene, and obtain location information for each role throughout the session using the role for each scene. there is. Here, the location information for each role may be a location distribution or average location of roles over the entire session.

Now, the analysis device 1400 may identify a formation based on location information for each role (S4700). For example, the controller 1440 inputs the location information of the role into a deep learning algorithm trained using training data labeled with a formation identifier recognizable by a worker in the relevant field to determine the formation of the corresponding session. can As another example, location information of roles obtained from different sessions may be clustered, a group including the location information of this session may be determined, and a formation identifier assigned to the group may be determined as a formation identifier for this session. A detailed description of this has already been described above.

Referring to FIG. 14 , an example of a method for providing tactical information for team sports according to an embodiment of the present disclosure includes, in the method according to FIG. 10 , acquiring location information for each role related to a formation period (S5600). , normalizing location information for each role using the role arrangement for the formation period (S5700), and identifying formations based on the normalized location information for each role (S5800). Hereinafter, each step of this example will be described.

In this example, when the formation period in which the analysis device 1400 divides the target session is obtained, location information for each role may be obtained (S5600), and location information for each role may be normalized using the role arrangement for the formation period. Yes (S5700). Here, normalization may be performed by reallocating roles according to location information for each role obtained in different formation periods. Given that the initial role assignments in the player-role assignments in this description are arbitrary, the player-role assignments defined in this session may be different from the player-role assignments defined in other sessions. That is, for example, when the player-role assignment is expressed as a role substitution sequence, since the order of vectors between different sessions may not match, it is necessary to normalize the order within vectors between similar roles. Compatibility between player-role assignments between different sessions (eg, different games) can be ensured by using location information and role arrangement for each role. This task may be performed by the controller 1440, and for this purpose, the analysis device 1400 clusters or classifies data on player-role assignments, role arrangements, and position distribution by role obtained from various sessions in advance. Through this, it is possible to use an algorithm that can match the order of players in the player-role assignment.

Now, the analysis device 1400 may identify a formation based on location information for each role (S5800). For example, the analysis device 1400 may identify a formation for a corresponding session using a deep learning algorithm from location information for each role or a classification algorithm obtained in a clustering process. Since a detailed description of this has been described above, it will be omitted here.

Referring to FIG. 15 , an example of a method for providing tactical information for team sports according to an embodiment of the present disclosure includes obtaining a player tracking data set (S6100) and obtaining a player-role assignment (S6200). ), obtaining a main player-role assignment (S6300), and identifying a role for each player based on the main player-role assignment (S6400). Hereinafter, each step of this example will be described. Hereinafter, each step of this example will be described.

In this example, the analysis device 1400 may obtain a player tracking data set (S6100), obtain a player-role assignment (S6200), and obtain a main player-role assignment (S6300). Here, the player tracking data set is related to the entire session or the formation period, and in the latter case, the role of each player maintained constant during the formation period is obtained.

Next, the analysis device 1400 may identify a role for each player based on the main player-role assignment (S6400). Accordingly, the role of each player can be finally interpreted as a position name that can be identified by those working in the field. Specifically, the controller 1440 may determine the main player's main role according to the main player-role assignment, determine the location information of the role as the player's location information, and identify the player's position therefrom. In this case, a deep learning algorithm that receives position information of individual players or other players and outputs a position label may be used, or other classification algorithms may be used. In addition, in team sports having a more complex formation, the formation identifier according to the above example may be additionally used.

Referring to FIG. 16 , an example of a method for providing tactical information for team sports according to an embodiment of the present disclosure includes obtaining an assistant player role assignment in the method according to FIG. 15 (S7400), and a main player- A step of obtaining role change information in consideration of role assignment and sub-player role assignment (S7500) may be included. Hereinafter, each step of this example will be described.

In this example, when the main player role assignment is completed according to the above-described several examples, the analysis device 1400 obtains the sub player role assignment (S7400), and the role is switched in consideration of the main player-role assignment and the sub player role assignment. Information can be obtained (S7500). Here, the secondary player role may be a player-role assignment with a high frequency except for the main player role. The controller 1440 may detect the sub player-role assignment for the session by applying the above-described sub player role determination condition.

Next, the analysis device 1400 may acquire role change information in consideration of the main player-role assignment and the secondary player-role assignment. For example, controller 1440 may determine how often two player-role assignments occur based on the frequency of primary and secondary player-role assignments, or in primary and secondary player-role assignments. In consideration of the changed role of each player, roles or players whose roles are to be switched can be determined. In addition, a detailed description of the role change information has been described above.

Referring to FIG. 17 , an example of a method for providing tactical information on team sports according to an embodiment of the present disclosure includes obtaining an image of a target session (S9100) and a player tracking data set for the target session. It may include obtaining (S9200), acquiring tactical analysis information from the player tracking data set (S9300), and bookmarking the video timeline using the tactical analysis information (S9400). Hereinafter, each step of this example will be described.

First, the analysis device 1400 may obtain an image of the target session (S9100) and a player tracking data set (S9200). Here, the image is an image captured by an external camera and may be an image of a sports game or training, and the player tracking data set may be player location information for the same session.

Next, the analysis device 1400 may obtain tactical analysis information from the player tracking data set (S9300). Here, the tactical analysis information may include the above-described role switching information, whether or not it is an irregular situation, and information about a formation period or a role period. For example, the controller 1400 uses the information obtained from the above-described several examples (eg, the main player-role assignment and the secondary player-assignment comparison result) to determine when a position switch between players occurs or when a position switch occurs. The time period for which it is performed can be specified. For another example, the controller 1400 may specify the occurrence time or occurrence period of the irregular situation through detection of the irregular situation based on the switch rate. For another example, the controller 1400 may detect a formation change point or a role change point to specify a formation period or a role period.

When the tactical analysis information is acquired, the analysis device 1400 may perform bookmarking on the video timeline using the tactical analysis information. For example, the controller 1440 may bookmark time information related to a position switch, time information related to a role period, time information related to a formation period, and the like to a time line of a video. To this end, the analysis device 1400 may perform time synchronization between the video and the player tracking data set in advance. On the other hand, here, when bookmarking, information identifying the type of event that occurred (for example, "overlap of fullback", "forward-midfielder position swap", "corner kick", etc.) is not simply marked with a timestamp. can also be displayed together.

On the other hand, it should be noted in advance that the methods according to the embodiments of the present disclosure described above are described/shown as using some prior methods according to other embodiments of the present disclosure only for convenience of description.

Embodiments of the present disclosure described above may be provided as a computer software program that implements all or part of the above-described method or operation. For example, a computer software program may include program code provided in a form tangibly recorded on a machine-readable medium, and the program code may include instructions for executing steps or operations of a method according to an embodiment of the present disclosure. can include Such a computer program may be downloaded and installed from a network through a communication unit, and/or installed from a removable medium, and may be executed by a processor capable of reading the program code, so that the embodiments of the present description Followed steps or actions can be executed.

In addition, the embodiments of the present description described above may be provided in the form of a computer-readable recording medium on which all or part of the above-described software program is recorded. The type of recording medium may be a conventional memory device such as a floppy disk, a hard disk, a CD-ROM, or a memory card, or may be downloaded through the Internet or a computer communication network. You can take it and use it.

8. 결론8. Conclusion

An embodiment according to the present disclosure provides a changepoint detection framework, SportsCPD. SportsCPD can distinguish tactically intended changes in formations and roles in team sports from temporary changes. First, the temporary role position phase relationship and position transition can be expressed as a binary matrix and a sequence of permutations, respectively. Using this, a non-parametric change point detection algorithm (eg, discrete g-segment) can be applied to high-dimensional or non-Euclidean data with frequent repetitions to find formation and role assignment change points. Since the concept of formation and role is the most basic and intuitive way to express the tactics of a sports team, tracking and summarizing changes in the team's formation and role is valuable in itself to practitioners in the field. Furthermore, a switch pattern search or set-piece detection may be performed using additional information such as a temporary substitution sequence. Accordingly, embodiments of the present disclosure may be widely used in the field of team sports.

Hereinafter, the SportsCPD of the present description will be described in more detail with respect to embodiments used in the field of team sports. However, it should be noted that the scope of the technical idea of the present disclosure is not construed as being limited by the following examples, and the following examples are merely intended to illustrate at least a portion of the technical idea of the present disclosure by way of example.

선수 추적 데이터 획득Acquisition of player tracking data

18 illustrates an exemplary process for obtaining player tracking data that may be used by embodiments of the present disclosure. Hereinafter, with reference to FIG. 18, a process for obtaining player tracking data that can be used by embodiments of the present disclosure will be described in detail.

In this description, “player tracking data” may be used as a concept including a sequence of positional data of a corresponding player during a target time period to be analyzed. The target time period includes a plurality of viewpoints, position data including information on the position of the player at each of the plurality of viewpoints may be obtained, and the player tracking data may include position data at each of the plurality of viewpoints. can

18 illustrates an acquisition procedure for a sequence of location data for each of the players of a team participating in a soccer game, for example. A soccer game is played by including a total of 11 players 1810 in one team, and can be divided into 10

field players

1811 and 1 goalkeeper 1813.

To acquire location data for each of the players 1810, location information obtaining devices 1820 may be matched for each player 1810. The location information obtaining device 1820 may be, for example, the sensor apparatus 1200 as described with reference to FIG. 9 of the present description. For example, a GPS device is shown as the location information obtaining device 1820 of FIG. 18, but the apparatus 1820 for acquiring location information for each of the players 1810 of the present description is not limited to the GPS device, and examples Obtaining location information of each of the players 1810, such as, for example, either GPS, LPS (e.g., UWB or Bluetooth-based, etc.) or OTS (image analysis) as described above with respect to the sensor device 1200. Any technique for this may be applied. Hereinafter, for convenience of description, a location tracking procedure based on GPS technology will be described as an example.

Players 1810 can each be instructed to wear a corresponding GPS device 1820 and participate in the game. The GPS devices 1820 receive GPS signals from satellites so that each of the GPS devices 1820 can output a global position 1830 for a corresponding player. For example, the GPS device 1820 can be configured to output GPS data (latitude, longitude, altitude) at a sampling rate of 10 Hz, i.e., at 0.1 second intervals. Information output by the GPS device 1820 may be transmitted to a server. For example, the server may be an analysis device 1400 as described with reference to FIG. 9 .

Therefore, assuming that the time normally required from the start to the end of a soccer game, including intermissions, is 120 minutes, 72,000 (120 minutes X 60 seconds/minute X 10 Hz) location data are obtained for each player. It can be. Depending on the embodiment, the number of local locations 1840 or relative locations 1850 obtained from the above GPS location data may or may not be the same as the number of global locations 1830 .

The analysis device 1400 may receive the global location 1830 from the GPS device and convert it into a local location 1840 according to the arena coordinate system. The stadium coordinate system may refer to a two-dimensional coordinate system in which one of the corners of the stadium is the origin, and for example, the x-axis is the length of the stadium and the y-axis is the width of the stadium.

The analysis device 1400 can be configured to refer to the determined reference point based on the geographic location 1830 to calculate a position 1850 relative to each player's reference point. In a dynamic team sport such as a soccer game, each player's position is highly variable, and it may be difficult to derive a meaningful tactical change based on the actual position before conversion to the opponent's position. However, the relative position with respect to the reference point at each time point may have a relatively small amount of change, and more efficient tactical analysis may be performed through analysis of the relative position.

Here, the reference point may be, for example, an average position of the positions of the field players 1811, but is not limited thereto. Since the position of the goalkeeper 1813 shows relatively little change throughout the game, it may be advantageous to determine a reference point relative to the positions of the rest of the field players 1811. As the reference point, various representative values such as the center point as well as the average of the positions of the players 1810 may be used.

Since the reference point is calculated for each frame included in the target period, for example, the reference point can be calculated for each 72,000 frames for 120 minutes, and accordingly, the relative position 1850 of each player 1810 for each frame is calculated can do. The location data for each of the players used in the embodiments of the present disclosure may be, for example, relative location data 1850, but is not limited thereto.

As described above, by securing location data for each player for each time point within the target period to be analyzed, a player tracking data set including a sequence of position data of a corresponding player during the target period can be obtained. there is.

역할 할당 정보 획득Obtain role assignment information

19 is a schematic flowchart of a process for acquiring role assignment information according to an embodiment of the present disclosure, and FIG. 20 is a detailed flowchart of the non-regular role assignment process of FIG. 19 . Hereinafter, a process for obtaining role assignment information according to an embodiment of the present disclosure will be described in more detail with reference to FIGS. 19 and 20 .

Methods and/or processes in accordance with one embodiment of the present disclosure may be performed by a computing device. According to one aspect, the computing device may be, but is not limited to, the analysis device 1400 as described with reference to FIG. 9 , and any arithmetic device having a processor and a memory may be used.

In the present description, “role assignment information” may refer to information including role assignments determined for each of a plurality of time points within a target period to be analyzed. “Role assignment” may indicate a plurality of role indexes respectively assigned to a plurality of players at each point in time. For example, role indexes may be assigned to distinguish which role each player is playing at a given time, such as role A being assigned to the first player and role B being assigned to the second player. Assignment may refer to information on whether a role index is assigned to each of these players at each point in time.

Referring to FIG. 18 , the process of obtaining role assignment information according to an embodiment of the present disclosure includes dividing a target session into at least two time intervals (S1910), acquiring a player tracking data set (S1920), and Acquiring assignment information (S1930), determining main role assignment (S1940), determining irregular role assignment (S1950), acquiring spatial information set (S1960), updating role assignment information It may include one or more of (S1970) and determining whether an additional update is required (S1980). Hereinafter, each step of this example will be described.

As shown in FIG. 18 , the computing device may divide the target session into at least two time intervals (S1910). The division of the target session may be, for example, but not limited to, division of a formation section and/or a role section as described herein.

More specifically, the computing device may divide a target session for a match, for example of a team sport, into at least two time periods. Here, the at least two time periods may include a first time period and a second time period divided based on a change in the formation of a team participating in the team sports game or a change in the role of at least some of the plurality of players. The division of the time period may be performed, for example, by a point-of-change detection process as exemplified herein, but may include division of a period divided by any division criterion. Hereinafter, according to one aspect, the target period for role assignment may be, for example, any one of the first time period and the second time period divided as described above.

Referring back to FIG. 18 , next, the computing device may obtain a player tracking data set (S1920). More specifically, the computing device may obtain a plurality of player tracking data sets for a plurality of players, and each player tracking data set may include a sequence of position data of a corresponding player during a target time period. That is, information on positions of each of a plurality of players at each of a plurality of viewpoints included in the target period may be obtained. Note that the process for obtaining player tracking data according to an embodiment of the present disclosure described above with reference to FIG. 18 may be applied, but is not necessarily limited thereto. Further, according to one aspect, when the computing device divides the target session into at least two time intervals, player tracking data for a plurality of points of time included in the first time period or a plurality of points of time included in the second time period can be obtained.

Next, the computing device may acquire role assignment information (S1930). That is, the computing device may obtain role assignment information using a plurality of player tracking data sets. The role assignment information includes a plurality of role assignments for each of a plurality of times within the target period, and each role assignment may indicate a plurality of role indices assigned to a plurality of players at a corresponding time point.

More specifically, when player tracking data sets are acquired and the position of each player is identified, a role for each player can be assigned using this.

In this description, “role” may mean any one of a plurality of participant categories capable of matching player positions constituting a formation. That is, before determining which position among a plurality of positions, such as "MF" or "CF", it may be classified according to the role of a game participant that can be matched to any one of a plurality of positions. For these roles, for example, a role index such as "role A" may be assigned, and as will be described later in this description, which position each role index corresponds to, for example, an artificial neural network-based deep learning technique or automatically identified by computer-based techniques such as clustering techniques.

On the other hand, in the present description, "player" may mean an entity whose location is tracked by the process of acquiring player tracking data. For example, if there is no substitution of players throughout a game of team sports and there is no change of roles between players, “players” and “roles” may be matched one-to-one. However, in team sports, at least one player substitution usually occurs in a game, and position changes between field players may also occur. Thus, the sequence of location data for one “role” may be a mixture of at least some of the player tracking data for two or more “players”.

The role assignment is information indicating which role identifier is assigned to each of a plurality of players at each point in time. Assignment of a role to each of the players may be performed to perform optimal assignment based on the distribution of positions for each of a plurality of roles and the position of the player at each point in time, for example, the log probability density as described above in this description. A Hungarian algorithm, such as a base cost matrix, can be used.

For such role assignment, it is required to first obtain a location distribution for each role. The location distribution for each role may represent the location distribution of each role within the entire session (eg, 120 minute period) or a target period. The location distribution for each role can be understood to be included in the "spatial information set" hereinafter in this description, and the spatial information set includes each role at each of a plurality of time points included in the time period to be analyzed for each role. may include a sequence of location data of For example, if the target session is 120 minutes and location data is acquired at a sampling rate of 10 Hz, the spatial information set for each role may be a sequence of 72K locations of the corresponding role.

However, the step of acquiring the first role assignment information (S1930) after the player tracking data set is acquired (S1920) is performed in a situation where information on the location of the player is acquired but information on the location of the role is not secured. . Therefore, in the step of obtaining role assignment information (S1930), the position distribution for each initial role obtained by temporarily allocating one role to the position distribution for each player may be used.

Relatedly, FIG. 21 is an example of an initial role assignment table based on player tracking data, and FIG. 22 shows an exemplary initial role assignment result. Hereinafter, initial role assignment will be described with reference to FIGS. 21 and 22 .

In the step of acquiring the first role assignment information (S1930) after the player tracking data set is acquired (S1920), the player tracking data set for the position distribution of each player is obtained, but the spatial information set for the position distribution for each role is obtained it is not done That is, for example, the position of the first player (Player 1) at each viewpoint is known, but the position of the role A at each viewpoint may not be known at all.

For example, if role A is CF, and Player 1 plays the role of CF in the first half and Player 2 plays CF in the second half, the location distribution of CF is the location data of Player 1 in the first half (eg For example, it may be a set of Player 1's positions from the 1st to the 36,000th viewpoint) and Player 2's position data in the second half (Player 2's positions from the 36,001st to 72,000th viewpoints).

However, in the step of acquiring the first role assignment information (S1930) after the player tracking data set is acquired (S1920), since it is not known which player is performing the CF position at any point in time, hypothetically fixed for each player Roles can be temporarily assigned. For example, it can be assumed that Player 1 and Player 2 perform roles A and B respectively throughout the target session.

As shown in FIG. 22 , the first player's trajectory 2110 may include measured positions 2111 of the first player at each of a plurality of viewpoints, and the second player's trajectory 2120 may include a plurality of The measured positions 2121 of the second player at each of the viewpoints may be included. Here, if role A is temporarily assigned to the first player and role B is temporarily assigned to the second player, the position distribution 2115 for the first player is the position distribution 2115 of role A ), and the position distribution 2125 for the second player becomes the same as the position distribution 2125 for role B.

As shown in FIG. 21 , for example, if roles A to J are temporarily assigned to each of the first to tenth players, the same role assignment as the temporary assignment is performed at all times.

Accordingly, as shown in 22 , for each of the roles A to J, an initial position distribution 2105 for each role can be obtained.

When the position distribution for each initial role is obtained according to the provisional role assignment for the player tracking data set, as seen above, based on the player tracking data set and the position distribution for each initial role, for example, by performing the Hungarian algorithm, each frame is obtained. Role assignment information may be obtained by performing role assignment for each player. However, the algorithm for assigning roles to players is not limited to the Hungarian algorithm, and for example, role assignment to each of the players may be performed in a manner that minimizes costs for each of the plurality of players.

For convenience of explanation, for example, when assigning 10 roles to each of 10 players so as not to overlap each other, role assignment can be performed to minimize cost for each player. Here, the cost may be, for example, assigning a role to a player such that a sum of distances between a representative position for a specific role and a position of the player in a corresponding frame is minimized. A representative position for a specific role may be, for example, an average position over the entire target period of the specific role. That is, for example, in allocating 10 roles to 10 players, optimal assignment is made when the sum of the distances between the 10 players and the average positions of the 10 roles assigned to each of the players is minimized. can judge

Relatedly, FIG. 23 illustrates role assignment considering the distance between player positions and role positions. As shown in FIG. 23 , at the time of the first frame, for example, a time stamp of 0.1 sec, the first allocation example 2310-1 and the second allocation example 2310-2 can be reviewed.

In the first allocation example 2310-1, it is assumed that role A is assigned to the first player and role B is assigned to the second player, and in the second allocation example 2310-2, the first player is assigned role B. Assume that role B is assigned and role A is assigned to the second player.

In the first assignment example 2310-1, the distance between the representative position 2130 of role A and the position 2311 of the first player in the first frame, and the representative position 2140 of role B and the position 2311 in the first frame determines the cost for the first allocation example 2310-1 by calculating the sum of the distances between the positions 2312 of the second player of ) and the distance between the position of the second player 2312 in the first frame and the distance between the representative position 2140 of role B and the position of the first player 2311 in the first frame are calculated, A cost for allocation instance 2310-2 can be determined.

Since the cost in the first allocation example 2310-1 is less than the cost in the second allocation example 2310-2, in the first frame for the first player according to the first allocation example 2310-1 Role A is assigned, and role B may be assigned for the second player.

Although an example of assignment to the first player and the second player is shown in FIG. 23 for convenience of description, it has been described above that role assignment is performed for each of a plurality of players participating in team sports at that time. there is.

In addition, role assignment may be performed for each of a plurality of viewpoints included in the target period to be analyzed, that is, for each frame. 24A to 24C show changes in role assignment of each of a plurality of viewpoints.

24A shows role assignment 2320 for the first player and the second player in the second frame. In the second frame, the distance between the representative position 2130 of role A and the position of the first player 2321 in the second frame, and the representative position 2140 of role B and the position of the second player in the second frame Since the cost according to the sum of the distances between (2322) becomes the smallest, role A and role B can be assigned to the first player and the second player, respectively.

However, looking at the role assignment 2330 for the first player and the second player in the 1,234th frame shown in FIG. The cost according to the sum of the distance between the position 2332 of B and the distance between the representative position 2140 of role B and the position 2331 of the first player in the 1,234th frame is the smallest, so that role B is is assigned and role A may be assigned to the second player.

Referring again to FIG. 24C , the role assignments 2340 for the first player and the second player in frame 72,000 are shown. In the 72,000th frame, the distance between the representative position 2130 of role A and the position 1 of the first player 2341 in the 72,000th frame, and the representative position 2140 of role B and the position of the second player in the 72,000th frame ( 2342) becomes the smallest, and role A and role B can be assigned to the first player and the second player, respectively.

25 illustrates an example of a changed role at a time point of role assignment change. As described with reference to FIGS. 24A to 24C , looking at the role assignment results for the first player and the second player, at the time of 0.1 timestamp, role A is assigned to the first player and role B is assigned to the second player. was assigned, looking at the role assignment 2510 at the time stamp of 123.4, the role B is assigned to the first player, the role A is assigned to the second player, and then the role assignment at the time stamp of 7,200.0. Referring to step 2520, it may be returned that role A and role B are assigned to the first player and the second player, respectively.

26 is an example of overall role assignment including a role assignment change section. As shown in FIG. 26 , during the first period 2610, it may be a regular situation in which role A and role B are assigned to the first player and the second player, respectively. However, at a first point in time 2621, a position switch between the first player and the second player may occur. The position switch may include, for example, a situation in which the first player keeps the empty space of the second player due to overlapping of the second player. Therefore, during the second period 2620 from the first time point 2621 to the second time point 2623, the first player and the second player may exchange roles, and the first player is assigned role B and 2 Players may be assigned Role A (see 2625). At the second time point 2623, the position switch ends, and during the third time period 2630, it may correspond to a regular situation in which role A and role B are assigned to the first player and the second player, respectively.

As shown in FIG. Obtaining role assignment information (1930) can be accomplished. That is, information about which role index (eg, role A) is assigned to a certain player (eg, the first player) at each of a plurality of points in time within the target period can be obtained.

Referring back to FIG. 19 , the computing device may acquire a spatial information set (S1960). Specifically, the computing device may obtain a plurality of spatial information sets associated with a target period for each of a plurality of role indexes based on previously obtained role assignment information and player tracking data sets.

For example, the spatial information set for the role index of role A may include information on location data of role A at each of a plurality of viewpoints within the target period. Since the initial position distribution for each role used for obtaining the role assignment information (S1930) of FIG. 19 is an arbitrary role index assigned to the position distribution for each player, the initial position distribution for each role for one role identifier is for one player. Only location data for However, after role assignment is performed based on the position distribution for each initial role, for example, in the step of acquiring role assignment information (S1940) of FIG. 19, role B is assigned to the first player instead of role A, Positions of multiple players may be included in the spatial information set for one role identifier.

In relation to this, FIG. 27 shows the update of the spatial information set for each role through role assignment, and FIG. 28 is an example of a role assignment table showing the update of the spatial information set of FIG. 27 .

As shown in FIG. 27, in a state 2750 in which role identifiers are temporarily assigned to each of a plurality of players, the positions of role A are all composed of the positions of the first player indicated by black circles 2711. and the positions of role B are all composed of the positions of the second player indicated by black triangles 2721. Therefore, looking at the initial position distribution 2770 for each role, the initial position distribution 2115 for each role A is determined based on the position distribution of the first player, and the initial position distribution 2125 for each role B is determined based on the position distribution of the first player. 2 Determined based on the distribution of players' positions.

On the other hand, looking at 2760 after allocating roles to players for each time point by the role assignment information acquisition step (S1930) of FIG. 19, the positions of role A are the first player indicated by black circles 2711. A position to which role A is assigned to A and positions to which role A is assigned to the second player indicated by white triangles 2713 may be included together. Also, the positions of role B may include positions where role B is assigned to the second player, indicated by black triangles 2721, and positions where role B is assigned to the first player, indicated by white circles 2723. can Therefore, looking at the updated position distribution 2780 for each role, the position distribution 2715 for each role of role A shows a more concentrated form compared to the initial position distribution for each role 2115 for role A, and the position distribution for each role of role B The location distribution 2725 shows a more concentrated form than the initial location distribution 2125 for each role of role B.

In other words, when role assignment for each player is primarily completed for each frame based on the initial position distribution for each role, position distributions of roles may be updated again based on the role assignment. In the initial position distribution by role, the position distribution was obtained by assuming that the first player and the second player fixed role A and role B throughout the target period, respectively, but the role assignment for each player was completed for each frame After that, the location distribution for each role may be calculated again using the obtained role assignment and player tracking data.

For example, referring to the role assignment shown in FIG. 28 , it can be seen that role B is assigned to the first player and role A is assigned to the second player in the first time section 2810 in the sorting by players. Accordingly, in the sorting by role, the position of role A in the first time interval 2820 may be the position of the first player, and the position of role B may be the position of the second player.

More specifically, when looking at role A, the first player performs role A in frames 1 to 1,233 and 8,736 to 72,000, and the second player plays role A in frames 1,234 to 8,735. The position distribution for each role may include positions in frames 1 to 1,233 and 8,736 to 72,000 of the first player and positions in frames 1,234 to 8,735 of the second player.

As described above, the spatial information sets for each role obtained in the spatial information set acquisition step (S1960) of FIG. 19 may include a location distribution for each role reflecting the role assignment according to the role assignment information acquisition step (S1930). there is. The spatial information sets for each of the role indices obtained based on the role assignment results are more concentrated and better represent the location of the corresponding role compared to the initial location distribution for each role according to the temporary role assignment. can include

On the other hand, according to one aspect of the present disclosure, among role assignments for each of a plurality of views, a role assignment corresponding to noise is removed and a spatial information set for each role is obtained using only the remaining valid role assignments. can In other words, in updating the position distribution for each role in the step of acquiring the spatial information set (S1960), removing a noise element may be accompanied.

Position distribution by role is obtained to represent the positional characteristics of the role. For example, in team sports such as soccer and basketball, the movements of players are very fluid, and the team formation or player position participating in team sports This collapse can lead to a situation where the distinction is meaningless. The situation in which the formation or position collapses is representative of set piece situations such as corner kicks or free kicks. In this way, if the position distribution by role is updated by excluding the situation where the formation or position collapses as noise, the positional characteristics of the role can be more accurately reflected in the position distribution by role, and through this, tactical analysis on various team sports is performed. accuracy can be improved.

In particular, according to one aspect of the present disclosure, when a target session is divided into at least two time intervals (S1910) and spatial information sets for the divided periods are obtained (S1960), the need for noise removal and the effect of improving accuracy may appear larger. As exemplified above, assuming that the entire session for team sports is 120 minutes, if the analysis is performed on the entire session without a separate time period segmentation, for example, the analysis is performed including a set piece situation of approximately 10 minutes. However, it may not greatly affect the accuracy of analysis for a given role. However, when the target session is divided based on a predetermined criterion, such as a change point of a formation or a change point of a role, for example, one time interval may have a short time length of about 20 minutes. If repetitive set-piece situations occur continuously in a short time period of 20 minutes, set-piece situations of more than 10 minutes are included as noise situations, and the accuracy of tactical analysis based on position distribution by role for the entire period of 20 minutes can be very low. Therefore, according to one aspect of the present disclosure, when a target session is divided into at least two time intervals (S1910) and spatial information sets for the divided periods are acquired (S1960), the location distribution for each role is updated to avoid noise situations. It is required to remove

Noise to be removed may be referred to as a "non-regular situation" in this description. Relatedly, FIG. 29 is an example of a role assignment table for determining non-regular role assignment, and FIG. 30 is an example of switch rate determination according to the role assignment table of FIG. 29 . Hereinafter, determination of non-regular role assignment according to an embodiment of the present disclosure will be described with reference to FIGS. 29 and 30 .

As shown in FIG. 19 , prior to obtaining the spatial information set (S1960), the computing device determines the main role assignment (S1940) and determines the non-regular role assignment (S1950), excluding at least one non-regular role assignment. It may be configured to acquire a plurality of spatial information sets associated with target periods for each of a plurality of role indexes based on the remaining role assignment information and a portion of the player tracking data sets corresponding to the remaining role assignment information (S1960).

That is, when a plurality of role assignments for each of a plurality of viewpoints are acquired through the role assignment information acquisition step (S1930), the computing device may determine a main role assignment based on the plurality of role assignments (S1940). According to one aspect, the primary role assignment may be the most frequent role assignment among a plurality of role assignments. As exemplarily shown in FIG. 29 , as in the first time point 2910 , the role assignment in which roles A to J are assigned to the first to tenth players, respectively, is the highest within the entire time interval to be analyzed. Since it occurs with frequency, role assignment in which roles A to J are assigned to the first to tenth players, respectively, may be determined as the main role assignment. However, it should be understood that the determination of main role assignment is not limited to the most frequent role assignment, and various criteria for selecting main role assignments may be applied.

Subsequently, as shown in FIG. 19 , the computing device may determine at least one non-regular role assignment from a plurality of role assignments in consideration of the main role assignment (S1950).

As shown in more detail in FIG. 20 , the step of determining the non-regular role assignment (S1950) includes obtaining distance information between the main role assignment and each role assignment (S1951) and determining the non-regular role assignment based on the distance information. (S1953) may be included.

The computing device may obtain information about distances from a primary role assignment for each of the plurality of role assignments ( S1951 ). Here, the distance may reflect different degrees between each of the plurality of role assignments and the main role assignment. Also according to one aspect, the distance may be a switch rate from a primary role assignment for each of the plurality of role assignments.

More specifically, the computing device may determine a degree of difference from the main role assignment for each role assignment for each of a plurality of time points included in the target period. In the role assignments at the time points included in the second period 2920 of FIG. 29, the roles of the first player and the second player are role B and role A, respectively, unlike the main role assignment, and there is a relatively large difference from the main role assignment. may not be On the other hand, it can be seen that the role assignments of the time points included in the third period 2930 of FIG. 29 are significantly different from the main role assignment.

The degree of difference between the main role assignment and each role assignment can be expressed as a switch rate calculated based on the number of different roles. Referring to FIG. 30 , all times included in the second period 2920 may have a switch rate of 20% because the roles for the first player and the second player are different from the main role assignment. On the other hand, referring to the point of time included in the third period 2930, it can be confirmed that there are 7 or more players who have different roles from the main role assignment, so each has a switch rate of 70% or more.

Referring to FIG. 20 , the computing device may determine non-regular role assignment based on information about distances of each role assignment to the main role assignment (S1953). For example, if the switch rate is 70% or more, the corresponding frame may be regarded as a non-regular situation, and this may be excluded from updating the location distribution for each role. Referring to the example of FIG. 30 , since the role assignment corresponding to the times included in the second period 2920 has a switch rate of 20%, it is not included in the non-regular situation, but the times included in the third period 2930 Since all of the role assignments corresponding to have a switch rate of 70% or more, the corresponding role assignments are considered non-regular situations, and the position of the role according to the corresponding time point may not be considered when updating the position distribution for each role.

That is, referring again to FIG. 19 , the computing device determines a target for each of a plurality of role indexes based on the remaining role assignment information except for at least one non-regular role assignment and a portion of the player tracking data sets corresponding to the remaining role assignment information. A plurality of spatial information sets related to the period may be acquired (S1960). According to the remaining role assignment information except for the non-regular role assignment, it is possible to know which role index is assigned to which player at each time point corresponding to the regular situation, and through the player tracking data set, at the corresponding time point of the corresponding player Since the position of can be known, the position of each of the roles can be known at each viewpoint corresponding to a regular situation. Accordingly, the computing device may obtain a plurality of sets of spatial information for each role including a sequence of location data at respective points in time within a target period of each role.

FIG. 31 shows the update of the spatial information set for each role according to the removal of the irregular frame, and FIG. 32 exemplarily shows that the role assignment is changed at the same point in time according to the update of the spatial information set of FIG. 31 . Hereinafter, the effect of removing the irregular situation will be described with reference to FIGS. 31 and 32 .

As shown in FIG. 31, looking at the position distribution 3150 for each role before determining the non-regular frame, the positions of role A as described with reference to FIG. A position to which role A is assigned and positions to which role A is assigned to the second player indicated by white triangles 2713 may be included together. Also, the positions of role B may include positions where role B is assigned to the second player, indicated by black triangles 2721, and positions where role B is assigned to the first player, indicated by white circles 2723. can

Then, looking at 3160 after the determination of the irregular frame, the position corresponding to the noise among the positions for role A can be removed as indicated by black hearts 3111, and the position corresponding to the noise indicated by black crosses 3121 As such, among positions for role B, positions corresponding to noise may be removed. In FIG. 31, for convenience of description, it is shown that noise occurs at the time when player 1 performs roles A and B, but in fact, at the time of formation collapse corresponding to noise, not only roles A and B, but also any other role. It can also be determined by performing

As a result, looking at the position distribution by role 3170 from which noise has been removed, the position distribution by role 3115 of role A is the updated position distribution by role of FIG. 27 as well as the initial position distribution by role 2115 of role A. 2715, it shows a more concentrated form, and the position distribution 3125 for each role of role B is compared to the initial position distribution 2125 for role B as well as the updated position distribution 2725 for each role of FIG. appear to be more concentrated. Therefore, the location distribution for each role can better express the locational characteristics of the corresponding role.

Referring to FIG. 32 , for the same player position at the same time point, it can be confirmed that the role assignment according to the initial position distribution for each role and the role assignment based on the noise-removed position distribution for each role may be different from each other.

More specifically, looking at the role assignment 3280 according to the initial position distribution for each role, the role assignment will be performed based on the initial position distribution 2115 for role A and the initial position distribution for each role 2125 for role B. can The distance between the representative position 2130 of role A and the position of the first player 3211 in the 24,562 frame, and the distance between the representative position 2140 of role B and the position of the second player 3212 in the 24,562 frame Since the cost determined by calculating the sum of is the lowest, in the role assignment 3280 according to the initial position distribution for each role, role A is assigned to the first player and role B is assigned to the second player in the 24,562 frame. can

On the other hand, looking at the role assignment 3290 according to the position distribution for each role from which noise has been removed, the position distribution 3115 for each role updated by reflecting the noise of role A and the position for each role updated by reflecting the noise of role B Role assignment can be performed based on the distribution 3125. The distance between the representative position 3230 of role A and the position of the second player 3212 in the 24,562 frame, and the distance between the representative position 3240 of role B and the position of the first player 3211 in the 24,562 frame Since the cost determined by calculating the sum of is the lowest, in the role assignment 3290 according to the location distribution by role with noise removed, role B is assigned to the first player in the 24,562 frame, and role A is assigned to the second player. can be assigned.

In other words, different role assignments can be generated by using the position distribution for each role updated through noise removal for the same player in the same frame, which better reflects the positional distribution of the role to improve the accuracy of tactical analysis. can help you do that.

Referring back to FIG. 19 , the computing device may update role assignment information using a plurality of spatial information sets (S1970). That is, the computing device replays a role for each of a plurality of players using role assignment information obtained based on the initial position distribution for each role and spatial information sets for each of the noise-removed roles according to one aspect. By assigning, role assignment information can be updated. According to one aspect of the present description, determining the main role assignment (S1940), determining the non-regular role assignment (S1950), obtaining a spatial information set based on the remaining role assignment information excluding the non-regular role assignment (S1960), and updating the role assignment information (S1970) may be repeatedly performed until a predetermined condition is satisfied.

That is, role assignment and noise removal for players based on the position distribution by role for each role with improved accuracy, and acquisition of position distribution by role with further improved accuracy and player role assignment and noise removal, The process of obtaining a more updated position distribution by role and reassigning roles to each of the players accordingly can be repeated.

In this iterative process, different roles can be assigned to the same player in the same frame in the role assignment performed based on the changed position distribution for each role as described with reference to FIG. 32 . However, such a change in role assignment does not occur for all players in all frames, and the number of frames or players whose role assignment is changed can decrease as the update of role assignment and spatial information set are repeated. It is possible to reach a point where the assigned role for a player does not change even with further reassignment.

In this regard, as shown in FIG. 19 , the computing device may determine whether an additional update of the plurality of spatial information sets or role assignment information is required (S1980). If it is determined that additional update is required for spatial information sets or role assignment information according to a predetermined criterion, spatial information sets are obtained based on previously updated role assignment information, and role assignment for each of the players is performed again. Additional updates to role assignment information can be performed. When it is determined that additional update is not required for the spatial information sets or the role assignment information according to a predetermined criterion, the position distribution for each role is calculated based on the role assignment information according to the update of the last role assignment information, and finally the roles A spatial information set for each can be obtained.

According to one aspect of the present disclosure, the computing device may determine whether additional update is required based on whether at least a portion of the role assignment information is changed as a result of updating the previous role assignment information. For example, if the role assignment information is not changed at all by the step of updating the previous role assignment information, it is determined that the update procedure for the role assignment information has converged, and it is determined that additional update is not required, and the previous space The information set may be determined as the final spatial information set.

However, there may be cases where convergence does not occur despite repetitive role assignment. In this regard, according to one aspect of the present disclosure, the computing device requests the additional update based on whether points in time at which the role assignment of each point in time is changed by the step of updating the previous role assignment information are equal to or less than a predetermined threshold value. can decide whether or not For example, if, among a plurality of points in time included in the target period, the point in time at which a change occurred in the role assignment before the update by the step of updating the immediately preceding role assignment information is equal to or greater than a predetermined number, it may be determined that additional update is still necessary. can For example, if the entire sequence for the target session includes 72,000 views, it may be determined to perform additional updates until half role assignments are changed at less than 1% of the total views.

According to another aspect of the present disclosure, the computing device can determine whether additional updates are required based on whether a predetermined number of updates have been achieved. For example, when renewal of role assignment is completed 10 times, the role assignment process may be terminated without further updating.

In addition, the computing device may define conditions or times of various repetition counts and repeat the update until the conditions or counts are satisfied.

When it is determined that additional update is no longer necessary, a space including a sequence of positional data at each of a plurality of views for each of all roles, in which role assignment information for each of the players for all views is finally obtained. Information sets can be obtained.

Meanwhile, an apparatus for providing tactical information on team sports according to an aspect of the present disclosure includes a processor and a memory, and the processor includes a plurality of player tracking data sets for a plurality of players - each player tracking data set, obtain - including a sequence of positional data of the corresponding player during the target period; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to - Obtain; determine a primary role assignment based on the plurality of role assignments; determine at least one non-canonical role assignment from the plurality of role assignments in view of the primary role assignment; A plurality of spatial information associated with the target period for each of the plurality of role indexes based on remaining role assignment information excluding the at least one non-regular role assignment and a portion of the player tracking data sets corresponding to the remaining role assignment information get sets; and updating the role assignment information using the plurality of spatial information sets; can be configured to A specific operation of the apparatus for providing tactical information on team sports according to an aspect of the present disclosure may follow the role assignment information acquisition process as described above.

포메이션 기간 분할Division of formation period

34 is a schematic flowchart of a formation time period division process according to an embodiment of the present disclosure, FIG. 35 is a detailed flowchart of obtaining spatial arrangement information for the changing point detection process of FIG. 34 , and FIG. It is a detailed flowchart of obtaining distance information for the change point detection process. Hereinafter, the division of the formation section according to an embodiment of the present disclosure will be described in more detail with reference to FIGS. 34 to 36 .

According to one embodiment of the present disclosure, for example, a target session for a team sports game may be divided into at least two formation time intervals based on a time point when a team formation of a team participating in a team sports game is changed. can Analysis of team formations participating in team sports is very important in the tactical analysis of team sports. In addition, in performing a more specific and high-level analysis, such as analysis of the role change of at least some players within the formation as well as analysis of the team formation, the point at which the team formation of a team participating in team sports is changed It can be helpful to secure more accurate and meaningful tactical information by identifying and dividing the formation section and performing high-level tactical analysis on each of the divided formation time sections.

Referring to FIG. 34 , the formation period division process according to an embodiment of the present disclosure includes obtaining a player tracking data set (S3410), detecting a change point for a formation change (S3420), and the detected change point It may include one or more of determining whether it is valid (S3430) or dividing the target session into time intervals (S3440). Hereinafter, each step of this example will be described.

As shown in FIG. 34 , the computing device may first obtain a player tracking data set (S3410). More specifically, the computing device may obtain a plurality of player tracking data sets for a plurality of players, and each player tracking data set may include a sequence of corresponding player's location data during a target session. A division target of the formation time period may be a target session corresponding to an operating time of a game of team sports. Information on positions of each of a plurality of players may be obtained for each of a plurality of viewpoints during the target session. Note that the process for obtaining player tracking data according to an embodiment of the present disclosure described above with reference to FIG. 18 may be applied, but is not necessarily limited thereto.

Next, as shown in FIG. 34 , the computing device may detect a change point for the formation change (S3420). More specifically, the computing device generates feature values corresponding to each of a plurality of time points within the target session based on the plurality of player tracking data sets obtained previously, and changes point detection (Change Point Detection) for these feature values. , CPD) algorithm can be applied. As a result, the computing device may detect a change point indicating a point in time at which a formation of a team participating in the target session is changed, among a plurality of points in time in the target session.

Dividing the target session into at least two formation time periods may be performed by detecting a change point, which is a specific time point at which a formation change is performed, among a plurality of time points included in the target session to be analyzed. Therefore, in order to detect a point of change, it may be required to generate a feature value capable of reflecting characteristics that may be different for each viewpoint for each of a plurality of viewpoints included in the target session. When feature values of each viewpoint are generated for each of a plurality of viewpoints, detection of change points based on feature values may be achieved by applying a predetermined change point detection algorithm to sequences of feature values.

Here, the change point detection algorithm may be a non-parametric change detection technique, for example, as described in detail in "3. Related Art" of this description, and more specifically, the change point detection algorithm is discrete g-division ( Discrete g-Segmentation) algorithm. However, as mentioned in "3. Related Art", the change point detection technique of the present disclosure is not limited to such a specific algorithm, and any change point of a characteristic can be detected based on a sequence of values to be judged. It will be appreciated that the CPD algorithm may be used.

According to one aspect of the present disclosure, the feature values for each time point for applying the change point detection algorithm are, for example, information about the position of the player or role itself at that time point, or the location between players or roles It can be created based on information about enemy relationships.

More specifically, each feature value for each of a plurality of viewpoints may include positions of a plurality of players at each viewpoint, player space arrangement at each viewpoint, positions of a plurality of roles at each viewpoint, or role space arrangement at each viewpoint. It may be generated based on at least one of

Since the player tracking data set has been previously obtained (S3410), information on positions of each of the plurality of players can be obtained at each point in time. Based on the positions of a plurality of players located at each viewpoint, a feature value representing a characteristic of a corresponding viewpoint may be calculated. For example, the average position of all players at a point in time may be determined as a feature value at that point in time. A point-of-change detection algorithm may be applied to a sequence of a plurality of points-in-time to detect a point at which the average position of all players significantly changes as a point of change representing a change in formation.

Meanwhile, a feature value indicating a characteristic of a corresponding view may be generated based on the player space arrangement at each view. The player space arrangement may reflect a positional relationship between a plurality of players at each point in time. The feature value based on the player space arrangement may include, for example, player adjacency information calculated based on adjacencies between players, as will be described later in this description.

On the other hand, based on the positions of the plurality of roles at each time point, a feature value representing a characteristic of a corresponding time point may be calculated, and for example, an average position of all roles at that time point may be determined as a feature value at that time point. .

In addition, a feature value indicating a characteristic of a corresponding view may be generated based on the role space arrangement at each view. The role space arrangement may reflect a positional relationship between a plurality of roles at each viewpoint. As will be described later in this description, the feature value based on the role space arrangement may include, for example, role adjacency information calculated based on adjacency between roles.

In the embodiments of the present description, the information used to generate the feature value of each viewpoint is not limited to one in terms of "player" or "role", and also in terms of "plural positions" or "space arrangement". not limited to either one.

However, in terms of "player" or "role", as described above, "player" and "role" may be different from each other, especially when a player substitution or directed positional change between players occurs within the target session; A "role" may reflect more changes to the positions that make up the formation.

Also, in terms of “multiple positions” and “spatial arrangement”, in dynamic team sports such as soccer or basketball for example, the positions of the players change very actively throughout the target session, regardless of the change of formation. Situations in which the positions of players or roles change may occur in various ways. On the other hand, the "spatial arrangement" reflecting the positional relationship between players or between roles may have much less change than the position of the players or roles themselves if the formations remain the same. Thus, "space placement" may be more advantageous for detecting changes in formation than the position itself.

Hereinafter, although feature values according to an embodiment of the present disclosure are not limited to being obtained based on predetermined ground data, a process of generating feature values of each viewpoint based on role space arrangement at each of a plurality of viewpoints. be described in more detail.

The step of detecting a change point for formation change based on the role space arrangement of FIG. 34 (S3420) includes obtaining role assignment information (S3421) and acquiring a spatial information set (S3423) as shown in FIG. 35. and acquiring spatial arrangement information (S3425).

First, the computing device may obtain role assignment information using a plurality of player tracking data sets (S4321). The role assignment information may include a plurality of role assignments for each of a plurality of time points in the target session, and each role assignment may indicate a plurality of role indices assigned to a plurality of players at a first corresponding time point. That is, the role assignment information may include a plurality of role assignments for each of a plurality of times, and the role assignment at each time point may include information about which role index is assigned to each of the plurality of players at that time point. there is. A procedure for generating role assignment information based on player tracking data sets may be at least a part of the process of "obtaining role assignment information" described above with reference to FIGS. 19 to 20 in this description.

Thereafter, the computing device may obtain a plurality of spatial information sets associated with a target session for each of a plurality of role indexes based on the role assignment information and the player tracking data sets (S4323). The spatial information set for each of the plurality of roles may include a sequence of role location data over a target period of the corresponding role. At least a part of the spatial information set acquisition procedure in the “role assignment information acquisition” process described above with reference to FIGS. 19 to 20 in this description may also be applied to the spatial information set acquisition procedure.

When spatial information sets are secured, the computing device may obtain role space arrangement information including a plurality of role space arrangements for each of a plurality of viewpoints (step S3425). Each role space arrangement may reflect a positional relationship between a plurality of role indices at a second corresponding viewpoint. That is, for each of a plurality of views included in the target period, role space arrangements that can indicate what kind of positional relationship the plurality of role indices have at each view can be obtained. Role space arrangement information may be understood as information including role space arrangements for a plurality of viewpoints.

In the following, an example of a role space arrangement will be described in more detail.

By acquiring the player tracking data set (S3410), role assignment information (S3421), and spatial information sets (S3423), the role assignment of each player can be known for each of a plurality of time points included in the target session. and since the location of the player can be known at each point in time, information on the location of all roles at each time point can be obtained.

A role space arrangement may be considered as a feature value for each time point for detecting a change point indicating a time point at which a formation is changed. In team sports, since formation can be expressed as a positional relationship between roles, the positional relationship between roles at each time point can be more advantageously used as a characteristic value at each time point for determining a time point for changing a formation.

For example, when 2, 5, and 3 players are placed at the forefront, center, and rear, respectively, this may be determined as a 3-5-2 formation. As described above, when considering position changes of roles in dynamic team sports, a positional relationship between roles rather than the pure position of a role per se may be suitable for detecting a change in formation.

Even if the positions of the roles change very dynamically in the position in the stadium, the relationship between a specific role and other roles adjacent to the role may not change significantly as long as the formation is maintained. Specifically, when looking at the center forward in the 3-5-2 formation, the distance between the side forwards and the center forward on both sides can change very flexibly, but when examining which position is adjacent to the center forward, even though it is Even if the position changes, in most cases it can be identified as a two-sided forward.

Accordingly, according to an embodiment of the present description, information on what adjacent roles are for each role may be used as location information between roles for each of a plurality of roles. That is, it is possible to generate feature values for each viewpoint based on the role space arrangement reflecting the positional relationship between a plurality of roles.

However, according to an embodiment of the present disclosure, location information between roles may be extracted based on adjacent role information for each role, but the technical idea of the present disclosure is not limited thereto, and for example, through the locations of 10 positions. Like generating feature values based on deep learning, it is also possible to use a field-based (x,y) coordinate system.

To explain in more detail the feature value generation based on the role space arrangement, the role space arrangement at each time point determines whether the first role among the plurality of roles is adjacent to the second role among the plurality of roles at each time point. Reflected role adjacency information may be included. That is, the role adjacency information may be information indicating whether at least one of the plurality of roles is adjacent to another one of the plurality of roles.

More specifically, the role adjacency information may include a role adjacency matrix reflecting whether each of the plurality of roles is adjacent to each other with respect to each of the plurality of roles. That is, with respect to all of the plurality of roles, whether or not they are adjacent to each other in a relationship between all other roles can be displayed as a matrix having rows for all roles and columns for all roles.

Here, the role adjacency matrix may be obtained based on performing Delaunay triangulation on a plurality of roles. That is, it is possible to determine which role is adjacent to which role using Delaunay triangulation.

Relatedly, FIG. 38 illustrates a Delaunay triangulation for role adjacency determination. The Delaunay triangulation may be dividing a polygon generated by connecting positions of each of a plurality of roles into triangles such that the total sum of the minimum angles becomes the largest. For example, as shown in FIG. 38, a hexagonal polygon is created by connecting positions (38a, 38b, 38c, 38d, 38e, 38f) of a plurality of roles, and the polygon is divided into triangles. can Comparing the first divided form 3810, the second divided form 3820 and the third divided form 3830, the sum of the minimum angles of the triangles in the first divided form 3810 is a1 + b1 + c1 + d1 and , the sum of minimum angles of triangles in the second division form 3820 is a2 + b2 + b2 + d2, and the sum of minimum angles of triangles in the third division form 3830 is a3 + b3 + c3 + d3. A1 + b1 + c1 + d1, which is the sum of the minimum angles of the triangles in the first division form 3810, is the largest, and the first division form 3810 can be determined as a division of triangles according to the Delaunay triangulation. Intuitively, dividing triangles into a form as close to an equilateral triangle as possible can be a form of Delaunay triangulation, and according to this, it is divided into a form with the smallest acute angle, and the further away the vertex is, the smaller the angle is. A result of connecting close points to each other can be derived. Accordingly, it is possible to reduce the possibility that each of the plurality of points is determined to be adjacent to a point having an excessively large separation distance in determining other points adjacent to each other.

Specifically, according to the Delaunay triangulation, points sharing a triangle can be determined to be adjacent to each other. For example, in the first division form 3810, points 38a, 38b, and 38f may be determined to be adjacent to each other, and points having an excessively large separation distance between points determined to be adjacent according to the remaining triangles are not found. However, for example, in the third partitioned form 3830, there is a

triangle including points

38a, 38c, and 38e, and there is a relatively large separation between

points

38a and 38c and also between

points

38a and 38e. A problem may occur in which points are judged to be adjacent to each other even though they have a distance. Therefore, the first division form 3810 according to the Delaunay triangulation can be effectively used to determine whether a plurality of points are adjacent to each other without a problem of having an excessively large separation distance.

In the method according to an embodiment of the present disclosure, Delaunay triangulation may be used in determining whether roles are adjacent to each other with respect to a positional relationship between a plurality of roles. Therefore, for example, when roles A to J exist, whether or not they are adjacent to each other can be determined for each of a plurality of roles. For example, as role B is determined to be adjacent to role A, role C, and role D, adjacent roles may be determined for each individual role.

As reviewed above, in one embodiment of the present description, the role space arrangement of each viewpoint reflects whether a first role among a plurality of roles is adjacent to a second role among a plurality of roles at each viewpoint. It may contain adjacency information. That is, the role adjacency information may be information indicating whether at least one of the plurality of roles is adjacent to another one of the plurality of roles, for example, including a result of the Delaunay triangulation.

Meanwhile, FIG. 39 is an example of a role adjacency matrix representing a positional relationship between roles. As described above, more specifically, the role adjacency information according to an embodiment of the present disclosure may include a role adjacency matrix reflecting whether each of a plurality of roles is adjacent to each other with respect to each of a plurality of roles. can That is, with respect to all of the plurality of roles, whether or not they are adjacent to each other in a relationship between all other roles can be displayed as a matrix having rows for all roles and columns for all roles. As exemplarily shown in FIG. 39 , a role adjacency matrix including all roles in each row and column and having a different value depending on whether the roles corresponding to each row and column are adjacent to each other at the intersection of the row and column can be obtained. . For example, it may be set to have a value of 1 if the roles are adjacent to each other and a value of 0 if the roles are not adjacent to each other, but is not limited thereto, and if an arbitrary value is assigned differently depending on whether the roles are adjacent to each other enough

Since feature values for each of a plurality of views are generated for each view, a role adjacency matrix as illustrated in FIG. 39 may be generated for each view.

Meanwhile, in the above, a procedure for generating feature values for each of a plurality of viewpoints based on the role space arrangement has been described, but the technical idea of the present disclosure is not limited thereto. For example, a player space arrangement may be used to generate feature values, and the player space arrangement at each time point is such that a first player among a plurality of players is adjacent to a second player among a plurality of players at each time point, and It may include player adjacency information reflecting whether or not there is a player. In addition, the player adjacency information may include a player adjacency matrix reflecting whether each of the plurality of players is adjacent to each of the plurality of players, for example, in a form similar to that shown in FIG. 39 . For example, a player adjacency matrix in which the adjacency of each of the first to tenth players is indicated by a specific value may be generated. Similar to the case of role adjacency, a player adjacency matrix may also be obtained based on performing Delaunay triangulation on the positions of a plurality of players.

Referring back to FIG. 34 , in the step of detecting the change point for the formation change (S3420), the computing device may detect the change point by applying a change point detection algorithm based on the calculated feature values for each point in time. there is.

Here, according to an embodiment of the present disclosure, the change point detection algorithm may be applied to information about distances between viewpoints. More specifically, the step of detecting the change point for the formation change (S3420) includes the step of acquiring information on the distance between characteristic values of viewpoints (S3427) and applying a change point detection algorithm to the information on the distance (S3429). Further steps may be included.

Relatedly, FIG. 36 is a detailed flow chart of obtaining distance information for the point of change detection process in FIG. 34 . Referring to FIG. 36 , the computing device may obtain information about a distance between a feature value of a first view and a feature value of a second view among feature values for each of a plurality of views (S3427).

Here, the distance between the feature values may be a value reflecting the degree of difference between the feature values. According to an embodiment of the present disclosure, if a case in which a feature value of each viewpoint is a role adjacency matrix is exemplified, a distance between feature values is, for example, a role adjacency matrix for a first viewpoint and a role adjacency matrix for a second viewpoint They can represent different degrees. According to one aspect, the distance value between the role adjacency matrices at each time point is the Manhattan distance obtained by summing the values for all rows and columns by calculating 0 if the values of the corresponding rows and columns of the two matrices are the same, and 1 if they are different. may be defined as

Relatedly, FIG. 40 is an exemplary diagram of a process of dividing a formation time period based on mutual distances of each of a plurality of frames according to an embodiment of the present disclosure. The example of the role adjacency matrix 4001 in the 1st frame of FIG. 40 is compared with the example 4002 of the role adjacency matrix in the 3rd frame, 0 is calculated if the corresponding row and column values are the same, and 1 is calculated if they are different for all rows. If the values are summed up for the column and the

resultant value

4003, 2 can be calculated. That is, in FIG. 40 , the Manhattan distance between the first frame and the third frame is 2, and according to one aspect of the present disclosure, the distance between the feature value of the first frame and the feature value of the third frame may be determined as 2.

Although the Manhattan distance has been described as an example, the distance between each matrix is not limited to the Manhattan distance, and a distance according to an arbitrary type such as a Hamming distance or an L2 distance is a distance between feature values. It should be understood that it can be used as

On the other hand, according to one aspect of the present description, the information on the distance between a plurality of feature values reflects the distance to each of the feature values of the plurality of views from each of the feature values of the plurality of views. may include an inter-frame distance matrix.

For example, for each of a plurality of viewpoints included in the target session, a matrix including all distance values between the plurality of viewpoints may be calculated based on all distances to each other viewpoint. That is, distances of all frames of all frames may be obtained and a matrix may be generated based on the distances, and such an inter-frame distance matrix may be used as information about distances between a plurality of feature values.

Referring to FIG. 40 , for example, when the distance value 4003 between frame 1 and frame 3 is calculated as 2, the first row 3 column and the third row 1 column of the inter-frame matrix 4004 have a value of 2. . In this way, by calculating all the distances between all frames and all frames, the inter-frame distance matrix 4004 can be completed.

Referring back to FIG. 36 , the computing device may detect a change point by applying a change point detection algorithm to information about a distance between feature values (S3429). As illustrated through FIG. 40 , when the information on the distance between feature values according to an embodiment of the present disclosure is an inter-frame distance matrix, the change point detection algorithm may be applied to such an inter-frame distance matrix. there is.

As shown in FIG. 40 , in a target session having 72,000 frames, for example, looking at division 4005 of an exemplary formation period, as a result of applying the change point detection algorithm, the change point can be determined at 31,000 frame time point, The target session may be divided into a first formation time interval of frames 1 to 31,000 and a second formation time interval of frames 31,001 to 72,000.

In the above, an inter-frame distance matrix reflecting the distance for each feature value for a plurality of views from each of the feature values for a plurality of views is exemplified as information on the distance between feature values, but the present description Information on the distance between feature values according to an embodiment of is not limited thereto.

For example, information about the distance between feature values according to an embodiment of the present disclosure includes time series distances reflecting the distance between a feature value for a target time point and a feature value for a previous time point of the target time point - time series distance Calculated for each of a plurality of viewpoints - may include a sequence of.

Relatedly, FIG. 41 is an illustration of a process for dividing a formation time period based on a sequence of distances between role adjacency matrices of each of adjacent frames according to one embodiment of the present disclosure.

As shown in FIG. 41, a sequence of distance values can be calculated from a sequence of distance values (time-series frames) of a frame-by-frame matrix. More specifically, the Manhattan distance between the example of the role adjacency matrix of the nth frame 4101 and the example of the role adjacency matrix of the n+1th frame 4102 can be calculated as 2, and thus the feature value for the nth frame and Information 4103 on the distance between feature values for the n+1th frame can be determined as 2. In this way, a distance between a feature value for a target view and a feature value for a view before the target view, that is, a distance between a feature value between two adjacent views is calculated for each of a plurality of views included in a plurality of target sessions, A sequence of feature value distances between viewpoints may be obtained. In this description, a sequence of distances between feature values between adjacent viewpoints may be referred to as a sequence of time-series distances.

The sequence of time series distances can be represented as in graph 4104 of FIG. 41 . It can be confirmed that the distance between feature values between adjacent views may be different for each view, and the fluctuating distance is displayed through a graph.

According to one aspect of the present disclosure, a change point detection algorithm may be applied to the sequence of time-series distances obtained as described above to detect a change point at a point in time when a formation is changed.

In review, according to the embodiments of the present disclosure, a positional relationship between roles is calculated for each frame, a difference (distance) in the positional relationship between each frame is calculated, and then information about the thus calculated distance is changed. By applying the point detection technique, it is possible to divide a sequence of viewpoints belonging to a target session to be analyzed into two.

That is, the information on the distance between feature values is not limited to the above examples, and according to one aspect of the present disclosure, the information on the distance between feature values to which the point of change detection algorithm is applied is at a reference point in time. It may include distances for each of the feature values corresponding to a plurality of viewpoints in the target session from the reference feature value for .

On the other hand, looking more closely at the change point detection based on the sequence 4104 of time series distances of FIG. 41 and the division into formation time periods of the target session, first, the sequence of time series distances is shown in one dimension in FIG. 41 . A changing point detection algorithm is applied to this sequence of time-series distances to divide the entire section into two periods, a first formation period 4110 and a second formation period 4120. Also, it is possible to divide the divided section into two again by repeating the same operation. For example, the second formation period 4120 is a 2-1 formation period 4121 by applying a change point detection algorithm to feature values of points in time included in the second formation period 4120 and performing period division. and 2-2 formation period 4122.

Referring back to FIG. 34 , the computing device may determine whether the detected change point is valid (S3430). Specifically, the computing device may determine whether the change point is valid based on characteristics of a period divided based on at least one detected change point. The computing device may check a predetermined division condition for the divided section based on the detected change point, determine division if the division condition is satisfied, and prevent division if the division condition is not satisfied.

According to one aspect of the present disclosure, the computing device may determine that the detected change point is valid in response to determining that the significance value (Scan Statics p) of the scan statistic according to the detected change point is equal to or less than a predetermined first threshold value. there is. For example, it may be required that the significance value of the segmentation-based scan statistic according to the change point be 0.01 or less. Here, the significance level may be understood as a value indicating the possibility that values within the division period are incorrectly included in the corresponding division, and as the value is lower, it may be understood that the division is appropriately performed.

Further, according to one aspect of the present disclosure, the computing device may determine that the point of change is valid in response to determining that the time lengths of both the divided first and second time periods are equal to or greater than a second predetermined threshold. there is. For example, it may be required that the lengths of both the first time period and the second time period, which are divided sections, be equal to or longer than 5 minutes. Since the formation change is unlikely to be repeated within a short period of time, for example, a division into intervals of 5 minutes or more can be determined as an effective division.

According to one aspect of the present disclosure, the computing device determines a difference between an average value of feature values for a plurality of points of time included in a divided first time period and an average value of feature values for a plurality of points of time included in a second time period. It may be determined that the change point is valid in response to a determination that is equal to or greater than a predetermined third threshold. For example, it may be requested that the distance between the average positional relationships (adjacency matrices) of the two divided sections be equal to or greater than 7.0. That is, it is possible to determine that the formations of the two sections are different from each other only when the evaluation value for the formation of the two divided sections is calculated and the difference is greater than the predetermined difference.

Meanwhile, according to one aspect of the present disclosure, the computing device satisfies all of the condition for the significance value of the scan statistic, the condition for the minimum length of time, and the condition for the difference between feature values of each section, as described above. It can be determined that detection of a valid change point or effective period division has been performed only when it is performed.

Referring back to FIG. 34 , the computing device may determine to divide the target session into time intervals in response to determining that the detected change point is valid ( S3440 ). That is, the computing device may divide the target session into at least two time periods based on the point of change, the at least two time periods comprising: a first time period in which a formation of a team participating in the target session is a first team formation; and It may include a second time period in which the team's formation is a second team formation.

Meanwhile, the division of the period to be analyzed into formation periods may be continuously performed until a division condition (eg, a condition for satisfying the validity of a detected change point) is no longer satisfied. For example, if a condition is not satisfied after dividing the entire session once, the section is not divided, so it may be determined that the formation is constant throughout the entire session. On the other hand, as described with reference to FIG. 41 , the entire section is divided into two periods of the first formation period 4110 and the second formation period 4120, and the characteristics of the times included in the second formation period 4120 Many such as dividing the second formation period 4120 into a second formation period 4121 and a second formation period 4122 by applying a change point detection algorithm to the values and performing period division. A case in which the target session is divided into sections of may also occur.

Meanwhile, an apparatus for providing tactical information on team sports according to an aspect of the present disclosure includes a processor and a memory, and the processor includes a plurality of player tracking data sets for a plurality of players - each player tracking data set, obtain - including a sequence of positional data of the corresponding player during the target session; By applying a change point detection algorithm to feature values for each of a plurality of time points generated based on the plurality of player tracking data sets, the time point at which the formation of a team participating in the target session is changed within the target session is reflected. detecting a change point; and at least two time periods in which a formation of a team participating in the target session is a first team formation and a first time period in which a formation of the team participates in the target session is a first team formation. 2nd Team Formation, including the 2nd Time Period - Split by; can be configured to A specific operation of the apparatus for providing tactical information for team sports according to an aspect of the present disclosure may follow the formation period division process as described above.

포메이션 식별, 포지션 식별 및 선수 식별자 매칭Formation Identification, Position Identification and Player Identifier Matching

37 is a schematic flow diagram of a matching process for formation identification, position identification and player identifier according to one embodiment of the present disclosure. Hereinafter, a matching process for formation identification, position identification, and player identifier according to an embodiment of the present disclosure will be described in more detail with reference to FIG. 37 .

Generally, "formation" refers to the tactical arrangement of players participating in a team sporting event. According to an embodiment of the present disclosure, a formation of a team participating in a corresponding game may be identified based on information about locations of a plurality of players and/or roles in a team sports match.

As discussed above, in the present description, “role” may mean any one of a plurality of participant categories capable of matching player positions constituting a formation. That is, prior to determining which position among a plurality of positions, such as "MF" or "CF", is classified by a role identifier according to the role of a game participant so that it can be matched with any one of a plurality of positions. it may have been According to an embodiment of the present disclosure, whether each role corresponds to a position may be automatically identified by computer-based technology.

On the other hand, in the present description, "player" may mean an entity whose location is tracked by the process of acquiring player tracking data. The sequence of location data for one “role” may be a mixture of at least some of the player tracking data for two or more “players”.

On the other hand, in this description, "player" is tracked as an entity participating in the game, but in practice, each player is identified as a first player or a second player to track the position of a distinct entity among a plurality of players. It does not directly match the player identifier, which contains actual information such as the player's name or jersey number. In particular, for example, in the case of using a wearable location information obtaining device as a method for tracking the location of a player, the acquired player tracking data may be strictly tracking information for each location information obtaining device. there is. In the case of using a location tracking device designated for each individual, identification information for each location tracking device and data on the personal information of the user of the device are obtained in advance to determine what personal information each player tracking data is about. may be identified, but in reality, location tracking devices are difficult to be fixed and used for each individual, and in many cases, game participants select and wear any one of the random location tracking devices for each game. According to the embodiment of the present disclosure, after determining the role of the player tracking device corresponding to each player tracking data set, identifying the corresponding position, and comparing it with information about players participating in the corresponding game, the player tracking data set is automatically determined. It is possible to match the player identifier corresponding to .

Referring to FIG. 37 , the formation identification, position identification, and matching process for player identifiers according to an embodiment of the present disclosure include dividing a target session into at least two time intervals (S3705), obtaining a player tracking data set (S3710) step, formation identification step (S3720) step, formation data acquisition step (S3730) step, player-position assignment step step (S3740) step step, player information step step obtaining (S3750) and each player It may include at least one of the steps of matching (S3760) with the player identifier. Hereinafter, each step of this example will be described.

First, as shown in FIG. 37 , the computing device may divide the target session into at least two time intervals (S3705). The division of the target session may be, for example, but not limited to, a division of formation period process as described herein. In addition, division of the target session is optional, and a matching process for formation identification, position identification, and player identifier according to an embodiment of the present disclosure may be performed on the entire non-divided target session.

More specifically, the computing device may divide a target session for a match, for example of a team sport, into at least two time periods. Here, the at least two time periods may include a first time period and a second time period divided based on a formation change of a team participating in a team sports game.

The division of the time period may be performed, for example, by a changepoint detection process as exemplified in the present description, but may include division of a period divided by any division criterion.

Referring back to FIG. 37 , the computing device may obtain a player tracking data set (S3710). More specifically, the computing device may obtain a plurality of player tracking data sets for a plurality of players, and each player tracking data set may include a sequence of corresponding player's location data during a target session. That is, information on positions of each of a plurality of players at each of a plurality of viewpoints included in the target session may be obtained. Note that the process for obtaining player tracking data according to an embodiment of the present disclosure described above with reference to FIG. 18 may be applied, but is not necessarily limited thereto.

Next, the computing device may identify a formation based on the player tracking data set (S3720). Specifically, the computing device may identify a formation of a team participating in a team sports game based on a formation structure for a target session generated based on a plurality of player tracking data sets. In the present description, “formation structure” may refer to information for determining a formation during a target period, which is an analysis target of a team participating in a team sports game.

Here, according to one aspect of the present disclosure, when the target session is divided into, for example, a first time period and a second time period, the step of identifying a formation (S3720) is based on the first formation structure for the first time period. identifying a first formation of the team for a first period of time with, and identifying a second formation of the team for a second period of time based on a second formation structure for a second period of time. can do.

In other words, according to one aspect of the present disclosure, a target session may be divided into at least two time intervals, for example a first time period and a second time period, and for each of the divided time intervals participation in a team sport The team's formation may be identified.

In team sports games, team formations are frequently changed more than once within a game according to tactical instructions or changes in circumstances. Therefore, when the team formation is changed, the target session for one game of team sports is divided into a plurality of formation periods, and the formation of each divided period is identified based on data corresponding to each of the divided periods, It can identify team formations with improved accuracy and can be used for tactical analysis.

Hereinafter, a “formation structure”, which is information or data for determining a formation during a target period, which is an analysis target of a team participating in a team sports game, will be described in more detail.

According to one aspect of the present disclosure, the formation structure may be created based on the positions of players or positional relationships between players. More specifically, the formation structure according to one aspect includes sequences of positions during a target session for each of a plurality of players, average positions during a target session for each of a plurality of players, and based on at least one of a sequence of player space placements, or an average value of player space placements for each point in time during the target session. Here, each player space arrangement may reflect a positional relationship between a plurality of players at each viewpoint.

According to another aspect of the present description, the formation structure may be created based on the location of roles or positional relationships between roles. More specifically, the formation structure according to one aspect includes sequences of positions during a target session for each of a plurality of roles, average positions during a target session for each of a plurality of roles, and a based on at least one of a sequence of role space placements, or an average of role space placements for each point in time during the target session. Here, each role space arrangement may reflect a positional relationship between a plurality of roles at each viewpoint.

In other words, the formation identification process according to an embodiment of the present disclosure may obtain data related to positions of players and/or roles during a target period for which a formation is to be identified, and may identify a formation of the corresponding period based on this. In addition, data related to a location that can be obtained by embodiments of the present disclosure include location data for each of a plurality of viewpoints for each of a plurality of entities. Accordingly, various combinations or modified forms of data on a plurality of positions of a plurality of entities at a plurality of viewpoints may be utilized for formation identification.

Hereinafter, exemplary methodologies for identifying a team formation for the analysis target period based on the formation structure will be described in detail.

First, according to one aspect of the present disclosure, formation identification may be performed using artificial neural network-based deep learning. In performing formation identification ( S3720 ), the computing device may determine a formation identifier corresponding to the previously obtained formation structure using an artificial neural network-based formation identification model. Here, the formation identification model is generated by learning an artificial neural network based on a plurality of training data sets, each training data set including a formation structure sample and a labeled formation identifier corresponding to the formation structure sample can

For example, when a formation structure corresponding to a target period to be identified is input to a formation identification model, the computing device may output a formation identifier corresponding thereto.

Relatedly, FIG. 42 shows exemplary labels of training data for artificial neural network-based formation identification according to an embodiment of the present disclosure. 42, for example, in a soccer game, team formation information 4230 is an array of numbers corresponding to the number of players such as “3-5-2” and “4-4-2”. is expressed A formation identifier 4220 may be designated for each formation information. For example, as shown in FIG. 42, a formation identifier of “1” may be designated for the “3-5-2” formation, and a formation identifier of “2” may be designated for the “4-4-2” formation. there is. Therefore, when the formation structure for the target period is input to the formation identification model and a formation identifier of “1” is output, it can be seen that the team formation of the target period corresponds to the “3-5-2” formation.

Referring to FIG. 42 , an exemplary training data set for a formation identification model according to an aspect of the present disclosure will be described. As shown in FIG. 42 , the training data set can include a formation structure sample 4210 and a labeled formation identifier 4220 corresponding to the formation structure sample. An input (ANN Input Feature) of a formation identifier model based on an artificial neural network according to an aspect of the present disclosure may be a formation structure, and an output (ANN Result label) may be a formation identifier for a corresponding formation structure. A training data set may be prepared by securing a plurality of sample formation structures, determining which formation each sample formation structure corresponds to, and labeling a formation identifier corresponding to the corresponding formation.

In FIG. 42 , it is illustratively shown that sets of respective representative positions for a plurality of roles for a particular frame are labeled with formation identifiers. According to one aspect, as shown in FIG. 42, a set of representative positions of a plurality of roles for one frame can be used as a formation structure. Here, one frame may be, for example, a representative time point of the target period (eg, any one of a middle time point, a start time point, or an end time point of the target period). Alternatively, the set of average positions of each role at a plurality of times in the target period may be used as a formation structure for the target period, and one of the three average positions of each role at a plurality of times in the target period The formation identifier of the label can be returned. In addition, a sequence of viewpoints including a set of positions of a plurality of roles may be used as a formation structure. For example, all position values of roles A to J for 10 frames are used as one formation structure, and one formation identifier may be labeled for this. The 10 frames may be sampled at predetermined intervals during the target period.

In other words, various types of data can be used as a formation structure in the case of generating a formation decision model by training an artificial neural network with a formation structure as an input and a formation identifier as an output.

For example, when formation identification is performed using deep learning (DL), formation identification may be exemplarily performed based on the location distribution of roles. Noise, etc. may be removed within a specific formation period, and for example, the sequences of positions for each of the 10 roles may be used as inputs to the artificial neural network, or the average positions within the formation period of each role, that is, the role The average position (x1, y1) of A or the average position (x10, y10) of role J may be used as an input to the artificial neural network.

According to another aspect of the present disclosure, formation identification may be performed by using an adjacency matrix reflecting a positional relationship between players or roles as an input of an artificial neural network.

In addition, when using an artificial neural network for formation identification, a sequence of adjacency matrices or an average adjacency matrix may be used as an input of the artificial neural network. Here, the average adjacency matrix may represent an average value matrix of adjacency matrices of all frames from which noise frames are removed within a formation period subject to formation identification. It should be noted that acquisition of a positional relationship or an adjacency matrix of roles may be performed by the embodiments described above in this description, but is not limited thereto.

Meanwhile, when data including values for a plurality of roles is used as a formation structure according to an embodiment of the present disclosure, alignment between values for roles included in the formation section of each formation period may be required. . For example, when a role adjacency matrix is used as a formation structure, rows and/or columns in a role adjacency matrix may be adjusted to correspond to rows and/or columns in another role adjacency matrix. That is, sorting can mean adjusting the columns and rows of another adjacency matrix for an arbitrary criterion adjacency matrix, which can be handled, for example, by reassigning the order of roles with a Hungarian algorithm based on the position distribution .

Relatedly, FIG. 43 is a conceptual diagram of a role alignment process for clustering between different formation periods or generating training data.

As described above in this description, “role” may not be the same as “position”, and a role index for each of a plurality of roles is first arbitrarily assigned. For example, role A in a specific target session and role A in another target session may be role indexes for different positions. For example, role A may be assigned to a role corresponding to CF's position in a specific target session or formation period, and role A may be assigned to a role corresponding to DF's position in another target session or formation period.

As exemplarily shown in FIG. 43 , for example, for the reference formation period 4301, the index of role A is the role corresponding to the right flank midfielder (RMF) position 4310 of the 3-4-3 formation. can be assigned to On the other hand, for the target formation period 4302, the index of role A is assigned to the role corresponding to the right center back (CB) position 4320 of the 3-4-3 formation, and the right flank midfielder (RMF) position ( 4310), the role B index may be assigned to the role.

Therefore, for a formation structure that includes information on a plurality of role identifiers, for example, the role identifier of the role corresponding to the right flank midfielder (RMF) position 4310 of the target formation period 4302 is assigned to role A. Like changing, a procedure for aligning a plurality of formation structures so that the same role index can be assigned to the same position may be required.

According to one embodiment of the present disclosure, each of the formation structure samples used for learning of the artificial neural network refers to each of a plurality of data elements included in the formation structure sample, respectively, and each of a plurality of reference data elements included in the formation structure. It may contain a plurality of normalized data elements ordered by assigning to. More specifically, by assigning role index A to role index J, which are data elements included in the formation structure sample, to role index A to role index J, which are reference data elements included in the reference formation structure, respectively, included in the formation structure sample The role indices may correspond to the same positions for each of a plurality of formation structure samples.

Here, the allocation may utilize, for example, the Hungarian algorithm as described above, but is not limited thereto. For example, a representative position is determined (reference position) for each of a plurality of roles included in the reference formation period 4301, and a representative position is also determined for each of a plurality of roles included in the target formation period 4302. After (target position), the assignment of the role index to the roles of the target formation period may be readjusted to minimize the cost of reference position-target position matching.

Meanwhile, according to another aspect of the present disclosure, formation identification may be performed using a clustering technique. For example, in performing formation identification (S3720), the computing device matches the formation structure to a first formation structure group, which is any one of a plurality of formation structure groups, and obtains a formation identifier corresponding to the first formation structure group. can be obtained

For example, if the formation identifier corresponding to the first formation structure group is "1" and the formation information corresponding to the formation identifier "1" is a "3-4-3" formation, the team formation for the target period is "3" You can tell it's a 4-3" formation.

Here, the plurality of formation structure groups may be generated by clustering a plurality of formation structure samples and labeling each cluster with a formation structure identifier.

Relatedly, FIG. 4 relates to an example of a formation group according to an embodiment of the present disclosure. As exemplarily shown in FIG. 4, a plurality of formation structures respectively corresponding to a plurality of formations include, for example, a first formation structure group 410, a second formation structure group 420, and a third formation structure group ( 430), a fourth formation structural group 440, a fifth formation structural group 450, and a sixth formation structural group 460. Each formation structure group may be labeled with a formation structure identifier of "1" to "6", for example.

In this regard, embodiments of the present disclosure may use artificial neural networks for formation identification, as well as various other computer-based techniques such as clustering algorithms.

When using a clustering algorithm for formation identification according to one aspect of the present disclosure, average adjacency matrices for each of a plurality of formation periods may be used. Specifically, clustering of a plurality of formation periods may be performed using a distance between average adjacency matrices for each of a plurality of formation periods. Here, the distance between the average adjacency matrices may be, for example, a Manhattan distance indicating the number of different values between the two matrices, but is not limited thereto.

Sample formation structures for each of a plurality of target sessions or sample formation structures for each of a plurality of formation periods are secured, and the secured plurality of sample formation structures are, for example, the first formation structure group 410 to the It can be clustered into 6 formation structural groups (460). For example, when the sample formation structure is an average adjacency matrix having an average value of adjacency matrices within a corresponding period, clustering may be performed based on a distance between average adjacency matrices for each period. Position identifiers may be labeled by determining which formation each of the clustered results, that is, for example, the first formation structure group 410 to the sixth formation structure group 460 corresponds to.

Subsequently, if it is determined which formation structure group the formation structure of the target period to be identified corresponds to, the formation of the target period may be identified based on the position identifier corresponding to the corresponding formation structure group.

Meanwhile, even in performing clustering, if the formation structure includes values related to roles, alignment of roles may be required. According to one aspect of the present disclosure, each formation structure sample includes a plurality of data elements arranged by assigning each of a plurality of data elements included in the formation structure sample to each of a plurality of reference data elements included in the reference formation structure. Can contain normalization data elements. As a procedure for aligning a specific role, a procedure described in relation to a training set of an artificial neural network may be borrowed.

Referring back to FIG. 37 , the computing device may acquire formation data (S3730). Specifically, when the formation for the target session or period is identified by the formation identification step ( S3720 ), the computing device may obtain formation data corresponding to the identified formation.

Here, the formation data may include information on a plurality of positions for the formation and position characteristic values for each of the plurality of positions. That is, the formation data may include information on which positions are included in a specific formation and information on whether each position has a positional characteristic or position distribution.

For example, formation information for the "3-4-3" formation is information on a plurality of positions, left wing (LW), center forward (CF), right wing (RW), left midfielder (LMF) ), 1st Central Midfielder (CMD 1), 2nd Central Midfielder (CMD 2), Right Side Midfielder (RMF), 1st Center Back (CB 1), 2nd Center Back (CB 2), 3rd Center Back ( It may include data indicating that CB 3) is included. In addition, the formation information for the “3-4-3” formation may include location characteristic values for each position as described above.

According to one aspect of the present description, the location characteristic value for each position may be generated based on at least one of a representative location for each position or a location distribution for each position. That is, it can be understood as a value reflecting whether or not each position has a certain characteristic positionally.

Referring back to FIG. 37 , the computing device may obtain a player-position assignment (S3740). Specifically, the computing device obtains player-position assignments for each of a plurality of players, wherein each player-position assignment includes position characteristic values for each of the plurality of players and a position for each of the plurality of positions. Based on the characteristic values, it may be generated by assigning each of a plurality of players to each of a plurality of positions. That is, it is possible to determine which player among a plurality of players of a team participating in team sports corresponds to which position.

Here, as described above, the location characteristic value for each position may be generated based on at least one of a representative location for each position or a location distribution for each position. That is, it can be understood as a value reflecting whether or not each position has a certain characteristic positionally. Also, the location characteristic value for each player may be generated based on at least one of a representative location for each player and a location distribution for each player. That is, it can be understood as a value reflecting whether or not each player has a positional characteristic.

In relation to this, information on a player's location may be obtained from player tracking data sets according to one aspect of the present disclosure, and information on a position location may be obtained from formation data according to one aspect of the present disclosure. Now that the information about the player and the position for each of the positions is obtained, by comparing them it is possible to obtain a player-position assignment that determines which player corresponds to which position.

Therefore, according to an embodiment of the present disclosure, if player tracking data sets including a sequence of player position data are obtained, the formation of a team participating in a team sports game is determined by processing the player tracking data sets. It is possible to identify and determine which position each player corresponding to the player tracking data set corresponds to.

Meanwhile, at least one player substitution may occur during a team sports game. Considering the case where a player replacement occurs, the player-position allocation as described above may be obtained based on data of an early period of a team sports game. For example, the initial period may refer to a first period when the target session is divided into a plurality of periods. Alternatively, the initial period may mean a period having a predetermined length of time (for example, 30 minutes) from the start of the team sports game.

Referring back to FIG. 37 , the computing device may obtain participant information (S3750). Specifically, the computing device may obtain player information of a team participating in the target session, where the player information includes a plurality of positions for a formation of a team participating in the target session, and each of the plurality of positions. It may include player identifiers corresponding to.

In relation to recent team sports games, player information of each team for each team sports game is submitted in advance in most cases. Here, player information may be referred to as various terms including, for example, roster, member list, or line-up.

The participating player information may include information about what kind of formation a team participating in the target session is. In addition, information about which positions are included in the corresponding formation may be included in the participating player information. In addition, information about which player takes each position and participates in the game is included, and this may be included as a player identifier. That is, information on the name and/or uniform number (uniform number) of a player participating in each position may be included in player participation information as a player identifier. As an example of a soccer game among team sports games, when a soccer game is broadcasted, the formation of the participating team is schematized and transmitted on the screen before the game starts, each position included in the formation is displayed, and the position is in charge It can be seen that the player identifier such as the uniform number, name, or photo of the player playing is displayed together. Display information of such relay broadcasting may be produced by player identification information previously submitted before the game.

Subsequently, as shown in FIG. 37 , the computing device may match each player with a player identifier (S3760). Specifically, the computing device may match each of the plurality of players for the target session to each of the plurality of player identifiers based on the player-position assignments. According to an embodiment of the present disclosure, a player tracking data set may be obtained, and the player tracking data set may be secured for each of a plurality of players. Also, according to an embodiment of the present disclosure, player-position assignments indicating which positions each of a plurality of players corresponding to each player tracking data set corresponds to can be secured. Using these player-position assignments and information on a plurality of positions and player identifiers included in player appearance information, each of a plurality of players corresponding to a plurality of player tracking data sets can be matched with each player identifier. there is.

Therefore, according to an embodiment of the present disclosure, even when a specific participant does not use the player positioning device, such as the sensor device 1200 of FIG. If the player tracking data set is obtained by classifying the participants, it is possible to automatically identify which name or uniform number the player is in the player tracking data set.

Meanwhile, an apparatus for providing tactical information on team sports according to an aspect of the present disclosure includes a processor and a memory, and the processor includes a plurality of player tracking data sets for a plurality of players - each player tracking data set, obtain - including a sequence of positional data of the corresponding player during the target session; And a formation structure for the target session generated based on the plurality of player tracking data sets, wherein the formation structure reflects information for determining a formation of a team participating in a team sports game during the target session. Based on identify the formation of the team by; can be configured to A specific operation of the apparatus for providing tactical information for team sports according to one aspect of the present disclosure may follow the formation identification, position identification, and player identifier matching processes as described above.

역할 공간 배치 기반의 포메이션 구조Formation structure based on role space arrangement

33 is a schematic flow diagram of a formation structure acquisition process according to an embodiment of the present disclosure. Referring to FIG. 33 , a formation structure acquisition based on role space arrangement according to an embodiment of the present disclosure and a formation determination process accordingly will be described.

As described above in connection with the "formation identification, position identification, and player identifier matching process" in this description, "formation structure" is information for determining the formation during the target period, which is the analysis target of teams participating in team sports games. can mean The formation identification process according to an embodiment of the present disclosure is to obtain data related to the location and/or positional relationship of players and/or roles during the target period for which the formation is to be identified, and to identify the formation of the corresponding period based on this. can

However, in terms of "player" or "role", as described above, "player" and "role" may be different from each other, especially when a player substitution or directed positional change between players occurs within the target session; "Role" may be more advantageous for the identification of formations, and even from the perspective of "position" and "positional relationship", considering the dynamic positional change of team sports, "positional relationship" may be more advantageous for identification of formations . In the following, an example in which the "formation structure" is obtained based on the role space arrangement reflecting the positional relationship between roles will be described.

As shown in FIG. 33 , the computing device may acquire a spatial information set of a role ( S3310 ). More specifically, the computing device may obtain a plurality of spatial information sets related to a target period for each of a plurality of roles based on a plurality of player tracking data sets of a plurality of players.

Here, each player tracking data set may include a sequence of positional data of a corresponding player during the target period. A plurality of player tracking data sets may be obtained by the process for obtaining player tracking data according to an embodiment of the present disclosure described above with reference to FIG. 18, but it should be noted that it is not necessarily limited thereto.

In addition, each spatial information set may include a sequence of location data of a corresponding role during a target period to be analyzed. Acquisition of spatial information sets may also be achieved by the above-described embodiments, but is not limited thereto.

Referring back to FIG. 33 , the computing device may acquire role space arrangement information ( S3320 ). The role space arrangement information includes a plurality of role space arrangements for each of a plurality of viewpoints within a target period to be analyzed, and each role space arrangement reflects a positional relationship between a plurality of roles at a first corresponding viewpoint. can do. The role space arrangement may be obtained, for example, according to an embodiment as described above with respect to the "formation period division process" of this disclosure, but is not limited thereto.

Subsequently, the computing device obtains a formation structure for a target period to be analyzed based on a plurality of role space arrangements (S3330), and the computing device may determine a team formation based on the formation structure (S3340). .

For example, as described above with respect to the "Formation Period Dividing Process" of this disclosure, each of the role space arrangements may cause a first role of the plurality of roles to be a second role of the plurality of roles at a first corresponding point in time. and may include role adjacency information reflecting whether or not they are adjacent to each other. Also, the role adjacency information may include a role adjacency matrix reflecting whether each of a plurality of roles is adjacent to each other with respect to each of the plurality of roles.

As described above, the role adjacency matrix may be obtained for each of a plurality of time points of the target period to be analyzed. As described above in connection with the "formation identification process" of the present description, in the identification of formation using computer-based technology, for example, the form of input data of a training data set for training an artificial neural network may be variously modified. Specifically, the formation structure may be a sequence of a plurality of role adjacency matrices for each of a plurality of viewpoints, or may be an average role adjacency matrix reflecting an average of a plurality of role adjacency matrices for each of a plurality of viewpoints.

Meanwhile, an apparatus for providing tactical information on team sports according to an aspect of the present disclosure includes a processor and a memory, and the processor includes a plurality of player tracking data sets for a plurality of players - each player tracking data set, a plurality of sets of spatial information associated with the target period for each of a plurality of roles based on a sequence of positional data of the corresponding player during the target period, each spatial information set comprising a corresponding sequence during the target period; obtains - including a sequence of location data of a role; role space arrangement information - the role space arrangement information includes a plurality of role space arrangements for each of a plurality of viewpoints within the target period, each role space arrangement comprising: a position between the plurality of roles at a first corresponding viewpoint; reflect enemy relationships - obtain; obtaining a formation structure for the target period based on the plurality of role space arrangements, wherein the formation structure reflects information for determining a formation of a team participating in a team sports game during the target period; and determining a formation of the team based on the formation structure; can be configured to

A specific operation of the apparatus for providing tactical information on team sports according to an aspect of the present disclosure may follow the formation structure acquisition process based on role space arrangement as described above.

역할 기간 분할Split role term

44 is a schematic flowchart of a role time period division process according to an embodiment of the present disclosure, and FIG. 45 is a detailed flowchart of obtaining distance information for the change point detection process of FIG. 44 . Hereinafter, role period division according to an embodiment of the present disclosure will be described in more detail with reference to FIGS. 44 and 45 .

According to one embodiment of the present disclosure, for example, a target period to be analyzed is divided into at least two or more role time periods based on a time point at which the role of at least some of a plurality of players participating in a team sports game is changed. can be divided Even when the formation of a team is maintained in team sports, players in charge of positions belonging to the formation may change each other. For example, while the "3-4-3" formation is continuously maintained, the player in charge of the role of "RMF" and the player in charge of the role of "CMF 1" play each other's roles according to tactical instructions. There may be cases where the roles are mutually exchanged to do so. According to an embodiment of the present disclosure, a time point at which the role of at least some of the plurality of players is changed may be detected, and the target period may be divided into role time periods based on the detected change point.

On the other hand, here, the change of role means "instructed role change" due to tactical necessity, etc., for example, overlapping of wingbacks and a cover play for it, which is distinguished from "temporary role exchange" in this description. has already been explained.

By dividing the role time period according to the indicated role change, high-level tactical analysis according to the role assignment within each role time period can be performed more accurately. In addition, the noise frame determination according to whether the role assignment is irregular can also be performed more accurately.

Referring to FIG. 44 , the role period division process according to an embodiment of the present disclosure includes obtaining a player tracking data set (S4410), detecting a change point for a role change (S4420), and detecting a change point One or more of determining whether it is valid (S4430) or dividing the target interval into role time intervals (S4440) may be included. Hereinafter, each step of this example will be described.

First, as shown in FIG. 44 , the computing device may divide the target session into at least two time intervals (S4405). The division of the target session may be, for example, but not limited to, a division of formation period process as described herein. In addition, division of the target session is optional, and the role period division process according to an embodiment of the present disclosure may be performed on the entire undivided target session.

More specifically, the computing device may divide a target session for a match, eg of a team sport, into at least two formation time periods. Here, the at least two time periods may include a first formation time period and a second formation time period divided based on a formation change of a team participating in a team sports game.

When a target session is divided into at least two formation time periods according to one aspect of the present disclosure, the target period subject to the role time period division process according to an embodiment of the present disclosure described below is, for example, a first It may be either a formation time period or a second formation time period.

As shown in FIG. 44 , the computing device may first obtain a player tracking data set (S4410). More specifically, the computing device may acquire a plurality of player tracking data sets for a plurality of players, and each player tracking data set includes position data of a corresponding player during a target period subject to role period division. Can contain sequences. Note that the process for obtaining player tracking data according to an embodiment of the present disclosure described above with reference to FIG. 18 may be applied, but is not necessarily limited thereto.

Next, as shown in FIG. 44 , the computing device may detect a change point for role change ( S4420 ). More specifically, the computing device generates feature values corresponding to each of a plurality of time points within the target period based on the plurality of player tracking data sets obtained previously, and changes point detection (Change Point Detection) for these feature values. , CPD) algorithm can be applied. As a result, the computing device may detect a change point indicating a time point at which the roles of at least some of the plurality of players participating in the team sports game during the target period are changed, among a plurality of points in time within the target period.

Dividing the target period into at least two role time periods involves detecting a change point, which is a specific point in time at which the roles of at least some of the plurality of players are changed, among a plurality of time points included in the target period to be analyzed. can be performed by Accordingly, in order to detect a point of change, it may be required to generate a feature value capable of reflecting a characteristic that may be different for each viewpoint for each of a plurality of viewpoints included in the target period. When feature values of each viewpoint are generated for each of a plurality of viewpoints, detection of change points based on feature values may be achieved by applying a predetermined change point detection algorithm to sequences of feature values.

The role period division process according to an embodiment of the present disclosure may be, for example, dividing each of the formation time periods into at least two or more role periods.

In relation to this, FIG. 46 is an exemplary diagram of obtaining a sequence of distances between role assignments between adjacent frames for role period division, and FIG. 47 is an exemplary diagram of role interval division using a change point detection algorithm. According to one aspect, the division into role time periods calculates a difference value sequence from a sequence of role assignments for each of a plurality of time points for a single formation period, as exemplarily shown in FIG. 46 , and It may be to apply a discrete g-division algorithm to the same difference value sequence to perform division into role time periods as exemplarily shown in FIG. 47 . It is as described above in the present description that a change point detection algorithm other than discrete g-segmentation can be applied.

Looking more specifically, each feature value for each of a plurality of viewpoints according to an aspect of the present disclosure may include role assignment of each viewpoint. As described in detail above in this disclosure, role assignment may indicate a plurality of role indices assigned to a plurality of players at a corresponding point in time. That is, a role index may be assigned to each player corresponding to a plurality of player tracking data sets, and the role assignment at a specific time point is information on which role indexes are assigned to each of the plurality of players at that time point can include

According to one aspect of the present description, when the characteristic value for each time point includes role assignment, the step of detecting a change point for a role change (S4420) is to obtain role assignment information using a plurality of player tracking data sets It may further include steps to do. The role assignment information may include a plurality of role assignments for each of a plurality of times within the target period. It should be noted that the above role assignment information may be acquired by the role assignment information acquisition process as described above in this description, but is not necessarily limited thereto.

Here, in the step of detecting a change point for a role change (S4420), the change point may be detected based on a distance between feature values between a plurality of viewpoints. As shown in FIG. 45 , the step of detecting a point of change (S4420) includes obtaining information on the distance between feature values of viewpoints (S4421) and applying a point of change detection algorithm to the information on the distance (S4423). steps may be included.

More specifically, the computing device may obtain information about a distance between a feature value of a first view and a feature value of a second view among feature values for each of a plurality of views (S4421). Here, the distance between the feature values may indicate different degrees between the feature values. For example, the distance between the first role assignment and the second role assignment may reflect the number of different roles between the first role assignment and the second role assignment.

According to one aspect, the information on the distance between feature values is a frame reflecting the distance to each of the feature values of a plurality of views from each of the feature values of a plurality of views included in the target period. -Can include an inter-distance matrix. Similar to obtaining an inter-frame distance matrix by calculating the distance between role adjacency matrices between all frames in the formation period division process according to the embodiment of the present disclosure, the distance between feature values between a plurality of views for role period division Also, a matrix including distance information of all viewpoints of all viewpoints may be included.

Alternatively, according to another aspect of the present invention, the information on distances between feature values includes a sequence of time-series distances reflecting the distance between a feature value for a target time point and a feature value for a time point previous to the target time point. can include Here, the time series distance may be calculated for each of a plurality of viewpoints.

For example, as shown in FIG. 46 , a feature value for each of a plurality of viewpoints that can be referred to for division into role time periods according to an embodiment of the present disclosure may be a role assignment at the corresponding viewpoint. there is. 46 illustrates role assignment for dividing a formation time period from 2,100.0 time stamp (frame 21,000) to 4,800.0 time stamp (frame 48,000) into a plurality of role times through formation period division, for example.

According to one aspect of the present disclosure, the information on distances between feature values may be a sequence of time-series distances reflecting distances between feature values of a target viewpoint and feature values of previous viewpoints of the target viewpoint. That is, a sequence of distances between feature values between adjacent viewpoints may be applied to the change detection algorithm.

As shown in FIG. 46, since there is no difference between the role assignment at time 2,100.0 and the role assignment at time 2,100.1, a distance 4610 between feature values between the two views becomes 0. On the other hand, for example, since the role assignment at time 3,044.2 differs from the role assignment at time 3,044.1 by 2, the distance 4610 between feature values between the two views becomes 2. In this way, a sequence in which distances between feature values of each viewpoint and previous viewpoints are calculated and such time-series distance values are listed for all viewpoints included in the target period may be applied to the change point detection algorithm.

Meanwhile, according to one aspect of the present disclosure, in the step of detecting a change point for the role change of FIG. 44 ( S4420 ), the computing device assigns at least one non-regular role among a plurality of role assignments for each of a plurality of viewpoints. Change points can be detected based on the remaining role assignments except for .

According to one aspect, the non-canonical role assignment comprises the computing device determining a primary role assignment based on a plurality of role assignments and determining at least one non-canonical role assignment from the plurality of role assignments in view of the primary role assignment. can be determined based on Also, the main role assignment may be the most frequent role assignment among a plurality of role assignments. More specifically, in determining the non-regular role assignment, it is possible to obtain information about distances from the main role assignment for each of the plurality of role assignments and determine the non-regular role assignment based on the information about the distances. there is. The distance between role assignments may represent different degrees between each of the plurality of role assignments and the main role assignment.

Determination of the non-regular role assignment may be determined using at least a portion of the procedures described in relation to the role assignment information acquisition process earlier in this disclosure. If the feature value of each time point used for role time period division is role assignment, and if the corresponding time point is a non-regular situation in which the formation collapses, excluding the role assignment at that time point and applying the change point detection algorithm to the role time period division Accuracy can be improved.

Next, referring again to FIG. 45 , the computing device applies a point-of-change detection algorithm to the information about the distance between feature values (S4423), so that the role of at least some of the players participating in the team sport is changed. A change point can be detected.

Meanwhile, referring to FIG. 44 again, the computing device may determine whether the detected change point is valid (S4430). Specifically, the computing device may determine whether the change point is valid based on characteristics of a period divided based on at least one detected change point. The computing device may check a predetermined division condition for the divided section based on the detected change point, determine division if the division condition is satisfied, and prevent division if the division condition is not satisfied.

Further, according to one aspect of the present disclosure, the computing device determines that the point of change is valid in response to determining that the time lengths of both the divided first role time period and the second role time period are equal to or greater than a second predetermined threshold. can do. For example, it may be required that the lengths of both the first role time period and the second role time period, which are divided sections, be equal to or longer than 5 minutes. Since there is little possibility that the role change is repeated within a short period of time, for example, a division into sections of 5 minutes or longer may be determined as an effective division.

Meanwhile, according to one aspect of the present disclosure, the computing device performs detection of a valid change point or division of an effective period only when both the condition for the significance value of the scan statistic and the condition for the minimum length of time are satisfied, as described above. can be determined to be

Referring back to FIG. 44 , the computing device may determine to divide the target period into role time intervals in response to determining that the detected change point is valid ( S4440 ). That is, the computing device may divide the target period into at least two role time periods based on the points of change, wherein the at least two time periods are roles of at least some of the plurality of players participating in the team sport during the target period. Based on the changed change point, a first role time period before the change point and a second role time period after the change point may be included.

Meanwhile, an apparatus for providing tactical information on team sports according to an aspect of the present disclosure includes a processor and a memory, and the processor includes a plurality of player tracking data sets for a plurality of players - each player tracking data set , containing a sequence of positional data of the corresponding player during the target period; A time point at which the role of at least some of the plurality of players is changed within the target period by applying a change point detection algorithm to feature values for each of a plurality of time points generated based on the plurality of player tracking data sets. detecting a point of change that reflects; and based on the point of change, the target period is defined as at least two time periods, wherein the at least two time periods include a first role time period before the point of change and a second role time period after the point of change. split into; can be configured to A specific operation of the apparatus for providing tactical information for team sports according to an aspect of the present disclosure may follow the role period division process as described above.

비정규 상황 검출 및 하이라이트 영상 데이터 추출Irregular situation detection and highlight video data extraction

48 is a schematic flowchart of a non-regular situation detection and highlight image data extraction process according to an embodiment of the present disclosure, and FIG. 8 is a diagram showing a relationship between a switch rate and set-piece occurrence according to an embodiment of the present disclosure. . Hereinafter, a process of detecting an irregular situation and extracting highlight image data according to an embodiment of the present disclosure will be described in more detail with reference to FIGS. 8 and 48 .

In this description, "irregular situation" may refer to a situation in which the formation or position of a team participating in a team sports game is disrupted, such as a set piece situation including a corner kick or a free kick. For example, in terms of tactical analysis based on role assignment for each of a plurality of players, such as a role assignment information acquisition process or a role period division process according to an embodiment of the present disclosure, data corresponding to an irregular situation may be removed as noise. there is. However, analysis of non-regular situations such as set pieces in tactical analysis of team sports games can have an important meaning in itself. Detection of irregular situations within a game can be used for statistical analysis of irregular situations, such as frequency of occurrence, time of occurrence, and duration of irregular situations in the corresponding game. Alternatively, strategies or tactics in non-regular situations, such as team set pieces participating in team sports, may also be an important analysis target. In addition, non-regular situations such as set piece situations are likely to correspond to highlights among game videos for team sports, so when detecting a point in time corresponding to non-regular situations, a basis for automatically extracting highlight videos from full-time game videos It can also be used as a resource.

According to one aspect of the present disclosure, the non-regular situation detection and highlight image data extraction process divides a target session for a game of a team sport into at least two role time periods, and for any one of the divided role time periods, the non-regular situation detection and highlight image data extraction process. Situation detection and highlight image data extraction may be performed. In addition, even when non-regular situation detection and highlight image data extraction are performed for the entire team sports game, it can be performed based on different role position distributions for each role time period. Since non-normal situation detection according to an aspect of the present disclosure may be performed based on role assignment, detection accuracy may be improved by applying different role location distributions for each role time period.

Referring to FIG. 48 , the process of detecting an irregular situation and extracting highlight image data according to an embodiment of the present disclosure includes acquiring a player tracking data set (S4810), acquiring role assignment information (S4820), and assigning a role. It may include at least one of detecting an irregular situation based on (S4830), acquiring full-time image data for a target section (S4840), and extracting highlight image data corresponding to an irregular situation (S4850). . Hereinafter, each step of this example will be described.

As shown in FIG. 48 , the computing device may obtain a player tracking data set (S4810). More specifically, the computing device may obtain a plurality of player tracking data sets for a plurality of players, and each player tracking data set may include a sequence of position data of a corresponding player during a target time period. That is, information on positions of each of a plurality of players at each of a plurality of viewpoints included in the target period may be obtained. Note that the process for obtaining player tracking data according to an embodiment of the present disclosure described above with reference to FIG. 18 may be applied, but is not necessarily limited thereto.

Next, the computing device may acquire role assignment information (S4820). That is, the computing device may obtain role assignment information using a plurality of player tracking data sets. Here, the role assignment information includes a plurality of role assignments for each of a plurality of times within the target period, and each role assignment may indicate a plurality of role indices assigned to a plurality of players at a corresponding time point. It should be noted that although the above role assignment information may be obtained according to at least some of the procedures described above in relation to the "role assignment information acquisition process" according to an embodiment of the present disclosure, it is not necessarily limited thereto.

Referring back to FIG. 48 , the computing device may detect an irregular situation based on role assignments ( S4830 ). That is, the computing device may detect an irregular situation within the target period based on the previously obtained plurality of role assignments. Here, the difference between the role assignment at the time corresponding to the irregular situation and the main role assignment may be greater than or equal to a predetermined threshold value. In other words, when the role assignments for each of the plurality of views are obtained, the computing device determines the main role assignment among them, calculates the difference between the role assignments corresponding to the respective views and the main role assignment, and calculates the difference value. A point of time equal to or greater than this predetermined threshold can be detected as an irregular situation.

Regarding the detection of irregular situations, FIG. 29 is an example of a role assignment table for determining irregular role assignment, and FIG. 30 is an example of determining a switch rate according to the role assignment table of FIG. 29 .

More specifically, when a plurality of role assignments for each of a plurality of viewpoints are obtained through the role assignment information acquisition step ( S4820 ), the computing device may determine a main role assignment based on the plurality of role assignments. According to one aspect, the primary role assignment may be the most frequent role assignment among a plurality of role assignments. As exemplarily shown in FIG. 29 , as in the first time point 2910 , the role assignment in which roles A to J are assigned to the first to tenth players, respectively, is the highest within the entire time interval to be analyzed. Since it occurs with frequency, role assignment in which roles A to J are assigned to the first to tenth players, respectively, may be determined as the main role assignment. However, it should be understood that the determination of main role assignment is not limited to the most frequent role assignment, and various criteria for selecting main role assignments may be applied.

The computing device may then consider the primary role assignment to determine at least one non-canonical role assignment from the plurality of role assignments. More specifically, determining the non-regular role assignment may include obtaining distance information between the main role assignment and each role assignment, and determining the non-regular role assignment based on the distance information.

The computing device may obtain information regarding distances from a primary role assignment for each of the plurality of role assignments. Here, the distance may reflect different degrees between each of the plurality of role assignments and the main role assignment. Also according to one aspect, the distance may be a switch rate from a primary role assignment for each of the plurality of role assignments.

The computing device may determine non-regular role assignment based on information about distances of each role assignment to the main role assignment (S1953). For example, if the switch rate is 70% or more, the corresponding frame may be regarded as an irregular situation. Referring to the example of FIG. 30 , since the role assignment corresponding to the times included in the second period 2920 has a switch rate of 20%, it is not included in the non-regular situation, but the times included in the third period 2930 Since all of the role assignments corresponding to have a switch rate of 70% or more, the corresponding role assignments can be regarded as non-regular situations.

As exemplarily shown in FIG. 8 , the detected irregular situations include an irregular situation (indicated by a box with a white background) relating to an offensive situation of a first team participating in a team sports game and an offensive situation of a second team. It can include non-normal situations (shown as boxes with black background).

Also, for example, a corner kick situation may be indicated by the letter "C", and a free kick situation may be indicated by the letter "F". Specifically, the detected irregular situations are the first team's free kick situation (811, 812, 813, 814), the first team's corner kick situation (821, 822, 823, 824, 825), and the second team's free kick situation (851 ), or at least one of the second team's corner kick situations (861, 862).

Among the irregular situations detected according to an embodiment of the present disclosure, classification information of irregular situations, such as which team each irregular situation relates to, or which situation of a corner kick or a free kick, may be determined by various criteria. there is. For example, the computing device may be configured to track the position of a ball during a team sports game, and may determine that a corner kick situation for the first team occurs when the ball is located near the corner of the first team's offensive position in an irregular situation.

Referring back to FIG. 48 , the computing device may acquire full-time image data for a target period (S4840) and extract highlight image data corresponding to a non-regular situation (S4850). That is, the computing device may obtain full-time image data for a target period and extract highlight image data including image data for a viewpoint corresponding to an irregular situation from the full-time image data.

For example, a set piece situation such as a corner kick or a free kick may be detected as an irregular situation, and such a set piece situation may correspond to a highlight, which is a key point in the entire game. Accordingly, when an irregular situation is detected and information on viewpoints corresponding to the irregular situation is secured, it is possible to know which viewpoint corresponds to a highlight among game images of the entire target session of a team sports game. Accordingly, it is possible to more easily extract a portion corresponding to a highlight image from the entire game image. According to one aspect, extraction of a highlight image may be performed automatically by a computing device without an editor's manual operation. In addition, according to another aspect, a highlight video time point may be displayed and provided as a bookmark in the time bar displayed together with the entire game video. For example, as shown in FIG. 8 , information about a set piece situation of a specific team may be displayed together on the time bar, and viewers watching the game video can conveniently view the specific team by selecting the corresponding bookmark. You can select and watch a video part for a specific set piece situation.

Meanwhile, an apparatus for providing tactical information on team sports according to an aspect of the present disclosure includes a processor and a memory, and the processor includes a plurality of player tracking data sets for a plurality of players - each player tracking data set , containing a sequence of positional data of the corresponding player during the target period; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to - Obtain; and detecting an irregular situation within the target period based on the plurality of role assignments, wherein a difference value between a role assignment at a time point corresponding to the irregular situation and a main role assignment is greater than or equal to a predetermined threshold value. A specific operation of the apparatus for providing tactical information on team sports according to an aspect of the present disclosure may follow the process of detecting an irregular situation and extracting highlight image data as described above.

플레이 스타일 분석 및 하이라이트 영상 데이터 추출Play style analysis and highlight video data extraction

49 is a schematic flowchart of a play style analysis and highlight image data extraction process according to an embodiment of the present disclosure, and FIG. 50 is a detailed flowchart of the play style information generation process of FIG. 49 . Hereinafter, a play style analysis and highlight image data extraction process according to an embodiment of the present disclosure will be described in more detail with reference to FIGS. 49 to 50 .

In the present description, "play style" may be understood to indicate a tendency or major strategy or tactics for playing a team sport of a team or player participating in a team sports game. For example, the play style for team sports may include information about a team formation that is mainly used, or information about the type or frequency of a partial tactic used. The information on partial tactics may mean a tendency to utilize switch play in which roles are temporarily exchanged between players while consistent formations and roles are maintained, for example, overlapping. Analysis of such a play style may occupy a very important part in performing a tactical analysis of an opposing team in team sports. By extracting the main tactics or strategies of the opposing team, counter tactics or strategies can be established in advance to prepare for these tactics or strategies, and players can become familiar with the tactics or strategies through actual training. On the other hand, for example, the timing of partial tactics such as switch play is likely to correspond to a highlight among game videos for team sports, so if the timing of partial tactics such as switch play is detected, the highlight video is automatically displayed in the full-time game video. It can also be used as evidence for extraction.

According to one aspect of the present disclosure, the play style analysis and highlight image data extraction process divides a target session for a game of team sports into at least two role time periods, and plays for any one of the divided role time periods. Style analysis and highlight image data extraction can be performed. In addition, even when performing play style analysis and highlight image data extraction for the entire team sports game, it can be performed based on different role position distributions for each role time period. Since the play style analysis according to one aspect of the present disclosure may be performed based on role assignment, analysis accuracy may be improved by applying different role location distributions for each role time period.

Referring to FIG. 49 , the play style analysis and highlight image data extraction process according to an embodiment of the present disclosure includes obtaining a player tracking data set (S4910), obtaining role assignment information (S4920), and role assignment It may include at least one of generating play style information based on (S4930), obtaining full-time video data for a target section (S4940), and extracting highlight video data corresponding to a switch play (S4950). there is. Hereinafter, each step of this example will be described.

As shown in FIG. 49 , the computing device may obtain a player tracking data set (S4910). More specifically, the computing device may obtain a plurality of player tracking data sets for a plurality of players, and each player tracking data set may include a sequence of position data of a corresponding player during a target time period. That is, information on positions of each of a plurality of players at each of a plurality of viewpoints included in the target period may be obtained. Note that the process for obtaining player tracking data according to an embodiment of the present disclosure described above with reference to FIG. 18 may be applied, but is not necessarily limited thereto.

Next, the computing device may obtain role assignment information (S4920). That is, the computing device may obtain role assignment information using a plurality of player tracking data sets. Here, the role assignment information includes a plurality of role assignments for each of a plurality of times within the target period, and each role assignment may indicate a plurality of role indices assigned to a plurality of players at a corresponding time point. It should be noted that although the above role assignment information may be obtained according to at least some of the procedures described above in relation to the "role assignment information acquisition process" according to an embodiment of the present disclosure, it is not necessarily limited thereto.

Referring back to FIG. 49 , the computing device may generate play style information based on role assignment (S4930). That is, information on a play style of at least one team or at least one player participating in a team sports match may be generated based on a plurality of role assignments.

According to one aspect of the present disclosure, the information on the play style may include information about the length of time of switch play reflecting the temporary role change by at least some of the plurality of players, information about the number of switch plays, or switch play may include at least one of information about the roles in which .

As described above, the play style according to an embodiment of the present description may include information about a partial strategy of a team or player participating in team sports. For example, analysis information on switch play indicating a temporary role exchange rather than a directed role change, such as wingback overlapping and winger cover play, may be included in the play style information. A multi-dimensional analysis of the switch play will be conducted, such as how often the switch play occurred during the target period, how long the switch play took place, and whether or not a switch occurred between players playing certain roles. can

Referring to FIG. 50 , more specifically, the step of generating play style information (S4930) includes a main role assignment decision (S4931), a sub role assignment decision (S4933), and switch play by comparing the main and sub role assignments. A step of obtaining information (S4935) may be included.

That is, the computing device may first determine a primary role assignment based on the previously obtained plurality of role assignments (S4931). Here, the main role assignment may be the most frequent role assignment among a plurality of role assignments. Then, the computing device may determine a sub-role assignment based on the plurality of role assignments (S4933). Here, the secondary role assignment may be the second most frequent role assignment among a plurality of role assignments. When the main role assignment and the sub-role assignment are determined, the computing device may compare the main role assignment and the sub-role assignment to obtain switch play information (S4935).

In this regard, referring to the role assignment table of FIG. 29 as an example, as in the first time point 2910, the total time that the role assignment in which roles A to J are assigned to the first to tenth players is an analysis target. Since it occurs with the highest frequency within the section, role assignment in which roles A to J are assigned to the first to tenth players, respectively, may be determined as the main role assignment.

Subsequently, role assignments may be sorted according to the frequency of role assignments, such as the second most frequent role assignment and the third most frequent role assignment among a plurality of role assignments as subrole assignments. However, it should be understood that the determination of main role assignment or sub-role assignment is not limited to the frequency of role assignment, and various selection criteria for main role assignment and sub-role assignment may be applied.

Relatedly, FIG. 51 is an example of frequency analysis of role assignment for play style information generation. As shown in FIG. 51 , role assignments in which roles A to J are assigned to the first to tenth players account for 67% of the total role assignments, and thus may be determined as the main role assignment 5110 . Then, in the main role assignment, the role assignments in which the role exchange between the first player and the second player occurred account for 15% and can be determined as the second most frequent role assignment 5120, and the eighth and ninth players in the main role assignment , role assignments in which temporary role exchanges have occurred for the role of the tenth player account for 13%, and thus may be determined as the third most frequent role assignment 5130 .

When the frequency analysis of the role assignments is completed in this way, referring to the second most frequent role assignment 5120, for example, the switch play between the player in charge of role A and the player in charge of role B is 15% in the entire target period. The fact that it occurs during the time corresponding to can be analyzed. Also, for example, among the second most frequent role assignments, role assignments for consecutive time points may be grouped and one group may be counted as one switch play occurrence. It is possible to analyze how many switch plays have occurred between the players in charge of role B. In this way, various analysis of various switch plays between players can be performed.

In relation to this, FIG. 6 is a graph of a role substitution sequence for each formation period according to an embodiment of the present disclosure, and FIG. 7 is a diagram illustrating an example of tactical analysis using a role substitution sequence according to an embodiment of the present disclosure.

6, for example, for each role interval of the first role interval 610, the second role interval 620, the third role interval 630, and the fourth to fifth role intervals 640, the time points at which the switch occurs The role assignments are sorted by frequency, and the duration corresponding to each role assignment is expressed in seconds. For example, in the first role section 610, for example, as shown in FIG. 7, when the roles are represented by numbers 1 to 10, the player of the number 4 role and the player of the number 6 perform the switch play. One period was the longest at over 60 seconds, followed by switch plays 3 and 5, switch plays 6 and 8 and 9, switch plays 1 and 10, and switch plays 5 and 6. You can check. In this way, it is possible to analyze how long the switch play between certain players lasted in a specific role section.

6 and 7 to review an analysis of exemplary play styles.

For example, in the first role period 610, the sum of the times corresponding to role assignments for the top three switch plays with high frequency corresponds to 140 seconds out of the entire 18 minutes of the role period, whereas in the third role period (630 ), the sum of the times corresponding to the role assignment for the top 3 switch plays with high frequency is only 35 seconds out of the total 13 minutes of the role section. Accordingly, it is possible to analyze that the players actively perform role switch in the first role period 610, but rarely perform switch play in the third role period.

On the other hand, switch plays 3 and 5 frequently occurred in the first role period 610, as shown in the first switch play example 710 of FIG. 7, to provide space for fellow players. It refers to a "false-9 play" in which the center forward falls behind the midfielder. However, it is possible to analyze that the players reduced the above switch play while the team formation changed to 4-3-3 during the second role period (620). In this way, by considering each role period and / or formation period together, and at least one or more of the identified formation, identified role, and information on the identified switch play, a higher-level tactical analysis of team sports can be performed. can

In addition, unlike in other role periods, it can be seen that in the first role period, the fullbacks (R1, R8) took on an attacking role along the side and actively played overlap with the wingers (R10, R9). As shown in FIG. 6 , in the first role period 610 , it can be confirmed that the switch play of (6 8 9) and the switch play of (1 10) are ranked high. This switch play is schematically represented in the second switch play example 720 of FIG. 7 . However, here, it is revealed that the overlap is different from each other on both sides. On the left side, R6 covers the position of R8, resulting in a cycle of 3 of (6 8 9), while on the right side, R4 does not cover the position of R1. A two-factor cycle of (1 10) occurred.

Meanwhile, as represented in the second switch play example 720 of FIG. 7 , two or more switches may be included together at one point in time. In view of this situation, according to one aspect of the present disclosure, the switch play is a first switch play reflecting a role shift between roles corresponding to the first role group and a role shift between roles corresponding to the second role group. , and at least one of the plurality of views may be reflected in both the first switch play and the second switch play. That is, for example, when the first switch play is a switch play of (6 8 9) and the second switch play is a switch play of (1 10), one A case may occur in which both the first switch play and the second switch play are included at the point of view. In analyzing the frequency of role assignments associated with a switch play according to one aspect of the present disclosure, the role assignment at a time point corresponding to the second switch play example 720 includes counting occurrences of the first switch play and second switch play It can be included in both counting for the occurrence of .

Meanwhile, referring to FIG. 6 again, in the second role period 620, it can be reviewed that a number of switch plays of

roles

6 and 9 have occurred, which is illustrated through the third switch play example 730 of FIG. As noted above, it can be analyzed that left winger R9 continuously attempted a cut-in play toward the penalty box through the switch (6 9).

In addition, in the 4th role period 640, after adding center forward R5, the team makes a general switch such as a switch between center forwards (3 5) or a switch between central midfielders (4 6) (eg within the same position). It is possible to analyze that the formation was kept constant except for the switch of ).

Referring back to FIG. 49 , the computing device may acquire full-time image data for the target period (S4940) and extract highlight image data corresponding to the switch play (S4950). That is, the computing device may obtain full-time video data for the target period and extract highlight video data including video data for a time point corresponding to the switch play from the full-time video data.

For example, a switch play situation in which at least some of a plurality of players participating in team sports temporarily exchange roles with each other, such as overlapping, may correspond to a highlight, which is a key point in the entire game. Accordingly, when the switch play situation is detected and information on points corresponding to the switch play is obtained, it is possible to know which point in time corresponds to a highlight among game images of the entire target session of the team sports game. Accordingly, it is possible to more easily extract a portion corresponding to a highlight image from the entire game image. According to one aspect, extraction of a highlight image may be performed automatically by a computing device without an editor's manual operation. In addition, according to another aspect, a highlight video time point may be displayed and provided as a bookmark in a time bar displayed together with the entire game video. For example, similar to the bookmarking of non-regular situations displayed on the time bar in FIG. 8, the switch play situation may also be displayed on the time bar as bookmarking (e.g., indicated by the letter "S"), and the game video may be displayed. By selecting the corresponding bookmark, the viewer can easily select and watch the video part about the switch play situation of a specific team.

An apparatus for providing tactical information for a team sport according to an aspect of the present disclosure includes a processor and a memory, wherein the processor includes a plurality of player tracking data sets for a plurality of players, each player tracking data set comprising a target obtains - including a sequence of positional data of the corresponding player over the period; Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to - Obtain; and generating information about a play style of at least one team or at least one player participating in a team sports match based on the plurality of role assignments; can be configured to A specific operation of the apparatus for providing tactical information on team sports according to an aspect of the present disclosure may follow the play style analysis and highlight image data extraction process as described above.

Since the embodiments disclosed in the present description are only illustrative of the technical idea of the present invention, the scope of the technical idea of the present invention is not limited to these embodiments. Therefore, the embodiments of the present disclosure described above are not only individual implementations and combinations, but also various modifications and variations that can be made to those skilled in the art without departing from the essential characteristics of the present invention. It will be understood that it belongs to the technical spirit of the present invention. Therefore, the protection scope of the present invention should be construed by the following claims, and all technical ideas within the equivalent range should be construed as being included in the scope of the present invention.

Claims

obtaining a plurality of player tracking data sets for a plurality of players, each player tracking data set including a sequence of positional data of a corresponding player during a target period;

Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to the - Obtaining;

determining a primary role assignment based on the plurality of role assignments;

determining at least one non-canonical role assignment from the plurality of role assignments in view of the primary role assignment;

A plurality of spatial information associated with the target period for each of the plurality of role indexes based on remaining role assignment information excluding the at least one non-regular role assignment and a portion of the player tracking data sets corresponding to the remaining role assignment information obtaining sets; and

updating the role assignment information using the plurality of spatial information sets; including,

How to provide tactical information for team sports.
According to claim 1,

The main role assignment is,

The most frequent role assignment among the plurality of role assignments,

How to provide tactical information for team sports.
According to claim 1,

The step of determining the non-regular role assignment,

obtaining information about distances from the primary role assignment for each of the plurality of role assignments, the distance reflecting different degrees between each of the plurality of role assignments and the primary role assignment; and

determining the non-canonical role assignment based on the information about the distances;

How to provide tactical information for team sports.
According to claim 3,

the distance,

Reflecting a switch rate from the primary role assignment for each of the plurality of role assignments,

How to provide tactical information for team sports.
According to claim 1,

determining whether an additional update is required for the plurality of spatial information sets or the role assignment information; Including more,

How to provide tactical information for team sports.
According to claim 5,

The step of determining whether the additional update is required,

determining whether the additional update is required based on whether at least a part of the role assignment information is changed by the updating of the role assignment information;

How to provide tactical information for team sports.
According to claim 5,

The step of determining whether the additional update is required,

Determining whether the additional update is required based on whether the time points at which the role assignment of each time point is changed by the updating of the role assignment information are equal to or less than a predetermined threshold value,

How to provide tactical information for team sports.
According to claim 5,

The step of determining whether the additional update is required,

determining whether the additional update is required based on whether a predetermined number of updates has been achieved;

How to provide tactical information for team sports.
According to claim 1,

A target session for a game of team sports is divided into at least two time periods, wherein the at least two time periods are divided based on a change in the formation of a team participating in the game or a change in the role of at least some of the plurality of players. comprising a first period of time and a second period of time;

wherein the target period is either the first period of time or the second period of time;

How to provide tactical information for team sports.
An apparatus for providing tactical information for team sports, said apparatus comprising a processor and a memory, said processor comprising:

obtaining a plurality of player tracking data sets for a plurality of players, each player tracking data set including a sequence of positional data of a corresponding player during a target period;

Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to - Obtain;

determine a primary role assignment based on the plurality of role assignments;

determine at least one non-canonical role assignment from the plurality of role assignments in view of the primary role assignment;

A plurality of spatial information associated with the target period for each of the plurality of role indexes based on remaining role assignment information excluding the at least one non-regular role assignment and a portion of the player tracking data sets corresponding to the remaining role assignment information get sets; and

updating the role assignment information using the plurality of spatial information sets; configured to

A device that provides tactical information for team sports.
A non-transitory computer-readable storage medium storing instructions executable by a processor, the instructions being executed by the processor to cause the processor to:

obtaining a plurality of player tracking data sets for a plurality of players, each player tracking data set including a sequence of positional data of a corresponding player during a target period;

Role assignment information using the plurality of player tracking data sets, wherein the role assignment information includes a plurality of role assignments for each of a plurality of time points within the target period, and each role assignment corresponds to the plurality of players at a corresponding time point. Indicating a plurality of role indices assigned to - Obtain;

determine a primary role assignment based on the plurality of role assignments;

determine at least one non-canonical role assignment from the plurality of role assignments in view of the primary role assignment;

A plurality of spatial information associated with the target period for each of the plurality of role indexes based on remaining role assignment information excluding the at least one non-regular role assignment and a portion of the player tracking data sets corresponding to the remaining role assignment information get sets; and

updating the role assignment information using the plurality of spatial information sets; configured to make

A computer readable storage medium.