KR20230033588A - A method for providing exercise load information - Google Patents

Abstract

A method for providing exercise load information for a target entity is provided. The method comprises the steps of: acquiring a target data set for the target entity during a target exercise session comprising a plurality of time units, wherein the target data set includes a sequence of kinematic information of the target entity for each of the plurality of time units and condition information reflecting characteristics associated with the target exercise session; and determining an estimated load index for the target entity from the target-specific information by using an artificial neural network, wherein the estimated load index reflects the exercise load of the target exercise session. According to the present invention, condition monitoring on a target entity can be done by outputting a notification message based on an estimated load index and a response load index for the target entity.

Description

How to provide exercise load information {A METHOD FOR PROVIDING EXERCISE LOAD INFORMATION}

The present disclosure relates generally to the field of sports analysis, and more specifically, without limitation, to techniques for calculating or using information reflecting an exercise load.

The importance of sports analysis is gradually increasing due to the explosive growth of the sports industry market and the development of sports science. In this trend, recently, electronic performance tracking systems (EPTS) that track sports players during games or training are being introduced one after another, focusing on major sports events including soccer.

The introduction of EPTS, which tracks the location or movement of sports players in real time, quantifies simple indicators such as distance traveled or average speed to relatively complex information such as heat maps for each player, enabling objective analysis of sports. It brings a new perspective to sports analysis, which used to rely on the observation and intuition of people like sports analysts.

However, these data-based sports analysis techniques have yet to deviate from the level of providing simple position or movement information or simply processed indicators, and research and development on analysis information or indicators from more diverse perspectives is underway. situation that is being demanded.

The following is a summary of specific embodiments disclosed in this disclosure. It should be understood that the aspects presented in the summary below are merely intended to provide a brief summary of specific embodiments and are not intended to limit the scope of the disclosure. Accordingly, it is disclosed in advance that the present description may include various aspects not presented below.

Inventive concepts disclosed herein and below provide methods, devices, and computer readable storage media for calculating or using information indicative of an exercise load.

One object of the present disclosure is to provide a method for determining exercise load information for a target entity using kinematic information of the target entity.

An object of the present disclosure is to provide a method for providing exercise load information, which outputs a notification message based on an estimated load index and a response load index for a target entity.

An object of the present disclosure is to provide a method for providing exercise load information to adaptively determine an estimated load index for a target entity based on at least one of identification information or characteristic information of the target entity.

An object of the present disclosure is to provide a method for providing exercise load information, which determines an estimated load index by considering both kinematic information about a target entity and state information reflecting characteristics of a corresponding exercise session.

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.

Embodiments include methods, devices, and computer readable storage media for calculating or using information indicative of an exercise load.

According to one embodiment, a method of providing exercise load information for a target entity, comprising a target data set for a target entity during a target exercise session comprising a plurality of time units, the target data set including the plurality of time units obtaining a sequence of kinematic information of the target entity for each of ; and determining an estimated load index reflecting an exercise load level of the target exercise session for the target entity, using an artificial neural network, from the target data set. Including, a method for providing exercise load information may be provided.

According to another embodiment, a method for providing exercise load information for a target entity includes, based on kinematic information of the target entity during the target exercise session, exercise of the target exercise session for the target entity. Calculating an estimated load index reflecting an estimated value for a load degree; Receiving a response load index reflecting a self-evaluation value for an exercise load degree of the target exercise session by the target entity; and outputting a notification message based on a comparison result between the estimated load index and the response load index. Including, a method for providing exercise load information may be provided.

According to another embodiment, a method of providing exercise load information for a target entity, comprising: acquiring target-specific information reflecting at least one of identification information and characteristic information for the target entity; a target data set for the target entity during a target workout session comprising a plurality of time units, the target data set comprising: a sequence of kinematic information of the target entity for each of the plurality of time units; including - obtaining ; and determining, from the target-specific information and the target data set, an estimated load index reflecting an exercise load degree of the target exercise session for the target entity, using an artificial neural network; Including, a method for providing exercise load information may be provided.

According to another embodiment, a method of providing exercise load information for a target entity, comprising a target data set for a target entity during a target exercise session comprising a plurality of time units, wherein the target data set includes the plurality of time units obtaining a sequence of kinematic information of the target entity for each of the target entity and condition information reflecting characteristics associated with the target exercise session; and determining an estimated load index reflecting an exercise load level of the target exercise session for the target entity, using an artificial neural network, from the target data set. Including, a method for providing exercise load information may be provided.

Exemplary embodiments other than the foregoing exemplary embodiments will be explained or made clear by the detailed description to be given later with respect to exemplary embodiments to be read in conjunction with the accompanying drawings.

The disclosed technology may have the following effects. However, it does not mean that a specific embodiment must include all of the following effects or only the following effects, so it should not be understood that the scope of rights of the disclosed technology is limited thereby.

According to embodiments of the present disclosure, it is possible to provide a quantified indicator of exercise load or improve management of a training session or sports player.

According to an embodiment of the present disclosure, it is possible to obtain information about the degree of exercise load of a target session for a target entity without going through a separate response procedure by determining exercise load information for the target entity using kinematic information of the target entity. can

According to an embodiment of the present disclosure, condition monitoring of the target entity may be performed by outputting a notification message based on the estimated load index and the response load index of the target entity.

According to an embodiment of the present disclosure, exercise load information is more accurately matched to the characteristics of a target entity by adaptively determining an estimated load index for a corresponding target entity based on at least one of identification information or characteristic information of the target entity. can decide

According to an embodiment of the present disclosure, by determining an estimated load index by considering both kinematic information on a target entity and state information reflecting characteristics of a corresponding exercise session, the condition of the target session for which the degree of exercise load is to be determined is reflected. More accurate exercise load information can be obtained.

The above summary is not intended to be an exhaustive list of all aspects of the invention. It is to be understood that the present invention includes all methods, apparatuses, and systems practicable from all suitable combinations of the various aspects set forth in the following detailed description and claims, as well as those summarized above.

1 illustrates a system for providing exercise load information according to an embodiment of the present disclosure.
2 is a block diagram of an attachable device according to an embodiment of the present disclosure.
3 is a block diagram of a server according to an embodiment of the disclosure.
4 illustrates an example of an exercise load index according to an embodiment of the present disclosure.
5 is a schematic flowchart of a method for providing exercise load information according to an embodiment of the present disclosure.
6 is a detailed flowchart of the step of determining the estimated load index in FIG. 5;
7 illustrates an exemplary artificial neural network used for calculation of an exercise load index according to an embodiment of the present disclosure.
8 illustrates an example of a training data set of an artificial neural network according to an embodiment of the present disclosure.
9 is a detailed view of an example of the kinematic information sequence of FIG. 8;
10 illustrates an example of an input data set of an artificial neural network according to an embodiment of the present disclosure.
11 illustrates another example of an input data set of an artificial neural network according to an embodiment of the present disclosure.
12 illustratively illustrates an entity-based coordinate system.
13 illustrates an arena-based coordinate system by way of example.
14 illustrates a difference between a motion direction of a previous view and a motion direction of a current view.
15 shows an example of kinematic information according to an embodiment of the present disclosure.
16 shows an example of kinematic information according to an embodiment of the present disclosure.
17 illustrates examples of coordinate systems capable of expressing kinematic information according to an embodiment of the present disclosure.
18 shows an example of static information.
19 shows an example of first kinematic information including speed-related information.
20 shows an example of second kinematic information including information about acceleration.
21 shows an example of third kinematic information including jerk-related information.
22 shows an example of each motion information based on an inertial coordinate system.
23 shows an example of each motion information based on a hybrid coordinate system.
24 is a schematic flowchart of a method for providing exercise load information according to an embodiment of the present disclosure.
FIG. 25 is a detailed flowchart of the estimated load index determining step of FIG. 24 .
26 illustrates activity information according to an embodiment of the present disclosure.
27 is an example of a wellness survey for securing daily readiness response information according to an embodiment of the present disclosure.
28 is a schematic flowchart of a method for providing exercise load information according to an embodiment of the present disclosure.
29 is a detailed flowchart of an embodiment of the step of determining the estimated load index of FIG. 28;
FIG. 30 is a detailed flowchart of another embodiment of the step of determining the estimated load index in FIG. 28;
31 is an exemplary diagram for determining a game load index.
32 is an exemplary view of determining an estimated load index using a game data set and a training data set.
33 illustrates an early fusion model in network fusion.
34 illustrates a middle fusion model in network fusion.
35 illustrates a late fusion model in network fusion.
36 is a conceptual diagram of a method for providing exercise load information for entity monitoring and early risk detection according to an embodiment of the present disclosure.
37 is a schematic flowchart of a method of providing exercise load information according to an embodiment of the present disclosure.
FIG. 38 is a detailed flowchart of the estimated load index calculation step of FIG. 37 .
39 is a detailed flowchart of the estimated load index determination step of FIG. 38;
40 illustrates a schematic structure of an artificial neural network for implementing a method for providing exercise load information according to an embodiment of the present disclosure.
41 illustrates an embodiment of providing exercise load information using kinematic information.
42 is an exemplary diagram of an artificial neural network structure considering time information.
43 is an exemplary view of a method of providing real-time exercise load information according to an aspect of the present disclosure.
FIG. 44 is an exemplary diagram of kinematic data up to time point N in FIG. 43 .
FIG. 45 is an exemplary diagram of kinematic data up to the point M in FIG. 43 .
46 illustrates a critical load warning procedure according to real-time exercise load measurement.

Since the present invention can make various changes and have various embodiments, specific embodiments are illustrated in the drawings and described in detail.

However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention.

Terms such as first and second may be used to describe various components, but the components should not be limited by the terms. These terms are only used for the purpose of distinguishing one component from another. For example, a first element may be termed a second element, and similarly, a second element may be termed a first element, without departing from the scope of the present invention. The terms and/or include any combination of a plurality of related recited items or any of a plurality of related recited items.

It is understood that when an element is referred to as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, but other elements may exist in the middle. It should be. On the other hand, when an element is referred to as “directly connected” or “directly connected” to another element, it should be understood that no other element exists in the middle.

Terms used in this application are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly dictates otherwise. In this application, the terms "include" or "have" are intended to designate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, but one or more other features It should be understood that the presence or addition of numbers, steps, operations, components, parts, or combinations thereof is not precluded.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related art, and unless explicitly defined in the present application, they should not be interpreted in an ideal or excessively formal meaning. don't

Hereinafter, with reference to the accompanying drawings, preferred embodiments of the present invention will be described in more detail. In order to facilitate overall understanding in the description of the present invention, the same reference numerals are used for the same components in the drawings, and redundant descriptions of the same components are omitted.

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.

A method for providing exercise load information for a target entity according to an aspect of the present disclosure includes a target data set for a target entity during a target exercise session including a plurality of time units, wherein the target data set includes the plurality of time units. obtaining a sequence of kinematic information of the target entity for each of the units; and determining an estimated load index reflecting an exercise load level of the target exercise session for the target entity, using an artificial neural network, from the target data set. can include

Here, the target entity is at least one sports team, and the method comprises performing a representative during the target workout session for the sports team based on estimated load indices for each of at least some athletes belonging to the sports team. Determining a load index; may further include.

Here, the step of determining the estimated load index comprises a plurality of training data sets including a first training data set, wherein the first training data set comprises a sequence of kinematic information of a first entity during a first exercise session. preparing the artificial neural network to be trained based on a score comprising and labeled with a score collected from the first entity for an exercise load of the first exercise session; inputting the target data set to an input layer of the artificial neural network; and obtaining the estimated load index based on a result value of an output layer of the artificial neural network. can include

Here, the method includes: measuring location information of the target entity for the target exercise session using a positioning device corresponding to the target entity; It further includes, and the kinematic information may be generated based on the location information.

Here, the kinematic information may be measured using an inertial sensor corresponding to the target entity.

Here, the kinematic information includes first kinematic information related to the velocity of the target entity; and second kinematic information related to the acceleration of the target entity.

Here, the kinematic information may further include third kinematic information related to the jerk of the target entity.

Here, the first kinematic information includes information on the magnitude of velocity in a target time unit; and information about an angular change amount between a direction of speed in a time unit previous to the target time unit and a direction of speed in the target time unit; can include

Here, the second kinematic information includes information on the magnitude of acceleration in a target time unit; and information about an angular change amount between a direction of acceleration in a time unit previous to the target time unit and a direction of acceleration in the target time unit; can include

Here, the second kinematic information includes information on the magnitude of acceleration in a target time unit; and information about an angular change amount between a direction of speed in a time unit previous to the target time unit and a direction of speed in the target time unit; can include

Here, the third kinematic information includes information on the magnitude of jerk in a target time unit; and information about an angular change amount between a direction of the jerk in a time unit previous to the target time unit and a direction of the jerk in the target time unit; can include

Here, the third kinematic information includes information on the magnitude of jerk in a target time unit; and information about an angular change amount between a direction of acceleration in a time unit previous to the target time unit and a direction of acceleration in the target time unit; can include

Here, the third kinematic information includes information on the magnitude of jerk in a target time unit; and information about an angular change amount between a direction of speed in a time unit previous to the target time unit and a direction of speed in the target time unit; can include

Here, the kinematic information includes a first axis for the length of a field in which an exercise session is performed; and a second axis for the width of the arena in which the exercise session is conducted; It can be expressed based on a field-based coordinate system including

Here, the kinematic information may include a forward and backward axis corresponding to the moving direction of the target entity; and a left and right axis corresponding to a lateral direction of the target entity; It can be expressed based on an entity-based coordinate system including

Here, the kinematic information may include information on at least one of a rotational movement in a roll direction, a rotational movement in a pitch direction, or a rotational movement in a yaw direction of the target entity; and information on at least one of angular velocity, angular acceleration, or angular jerk of the target entity.

Here, the estimated load index may be an estimated value of a rate of perceived fatigue (RPE) of the target entity for the target exercise session.

Here, the target data set may be configured to have a predetermined length by performing zero-padding prior to the kinematic information sequence.

An apparatus for providing exercise load information for a target entity according to an aspect of the present disclosure includes a processor and a memory, wherein the processor includes a target data set for the target entity during a target exercise session including a plurality of time units. obtain - the target data set comprising a sequence of kinematic information of the target entity for each of the plurality of time units; and determining an estimated load index reflecting an exercise load level of the target exercise session for the target entity, using an artificial neural network, from the target data set. 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 a target data set for a target entity during an exercise session, the target data set comprising a sequence of kinematic information of the target entity for each of the plurality of time units; and determining an estimated load index reflecting an exercise load level of the target exercise session for the target entity, using an artificial neural network, from the target data set. can be configured to do so.

A method for providing exercise load information on a target entity according to an aspect of the present disclosure includes, based on kinematic information on the target entity during the target exercise session, information about the target exercise session for the target entity. calculating an estimated load index reflecting an estimated value for an exercise load degree; Receiving a response load index reflecting a self-evaluation value for an exercise load degree of the target exercise session by the target entity; and outputting a notification message based on a comparison result between the estimated load index and the response load index. can include

Here, the calculating of the estimated load index may include a target data set for a target entity during the target exercise session including a plurality of time units, wherein the target data set includes the target data set for each of the plurality of time units. obtaining a - containing a sequence of kinematic information of an entity; and determining, from the target data set, an estimated load index for the target entity using an artificial neural network; can include

Here, the step of determining the estimated load index comprises a plurality of training data sets including a first training data set, wherein the first training data set comprises a sequence of kinematic information of a first entity during a first exercise session. preparing the artificial neural network to be trained based on a score comprising and labeled with a score collected from the first entity for an exercise load of the first exercise session; inputting the target data set to an input layer of the artificial neural network; and obtaining the estimated load index based on a result value of an output layer of the artificial neural network. can include

Here, the kinematic information includes first kinematic information related to the velocity of the target entity; and second kinematic information related to the acceleration of the target entity.

Here, the kinematic information may further include third kinematic information related to the jerk of the target entity.

Here, the estimated load index is an estimated rate of perceived fatigue (Rate of Perceived Exertion, RPE) of the target entity for the target exercise session, and the response load index is the perceived fatigue rate of the target entity for the target exercise session. can reflect

Here, the target entity is at least one sports team, and each of the estimated load index and the response load index may be a representative index determined based on indices for each of at least some athletes belonging to the sports team. .

Here, the target entity may be one sports player.

Here, the outputting of the notification message may include outputting the notification message in response to determining that the difference between the estimated load index and the response load index is greater than or equal to a first predetermined threshold value.

Here, the outputting of the notification message may include the outputting of the notification message in response to determining that a difference between the estimated load index and the response load index is greater than or equal to a predetermined second threshold value for exercise sessions that have been continued for more than a predetermined first threshold number of times. It can be configured to output a notification message.

Here, in the step of outputting the notification message, the ratio of second exercise sessions in which the difference between the estimated load index and the response load index is equal to or greater than a predetermined third threshold value among a plurality of consecutive exercise sessions is determined by the first It may be configured to output the notification message in response to a determination that the threshold ratio is greater than or equal to.

Here, the notification message may include an injury risk warning message for the target entity.

Here, the notification message may include an abnormal condition warning message for the target entity.

Here, the notification message may include a coach interview request message for the target entity.

Here, the notification message may include a type change recommendation message for the artificial neural network.

An apparatus for providing exercise load information for a target entity according to an aspect of the present disclosure includes a processor and a memory, wherein the processor is configured to, based on kinematic information for the target entity during a target exercise session, , calculating an estimated load index reflecting an estimated value of an exercise load degree of the target exercise session for the target entity; receive a response load index reflecting a self-evaluation value for an exercise load degree of the target exercise session by the target entity; and outputting a notification message based on a comparison result between the estimated load index and the response load index. 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, the instructions being executed by the processor to cause the processor to: calculating an estimated load index reflecting an estimated value of an exercise load degree of the target exercise session for the target entity, based on kinematic information for the target entity; receive a response load index reflecting a self-evaluation value for an exercise load degree of the target exercise session by the target entity; and outputting a notification message based on a comparison result between the estimated load index and the response load index. can be configured to do so.

A method for providing exercise load information on a target entity according to an aspect of the present disclosure includes obtaining target-specific information reflecting at least one of identification information and characteristic information on the target entity. ; a target data set for the target entity during a target workout session comprising a plurality of time units, the target data set comprising: a sequence of kinematic information of the target entity for each of the plurality of time units; including - obtaining ; and determining, from the target-specific information and the target data set, an estimated load index reflecting an exercise load degree of the target exercise session for the target entity, using an artificial neural network; can include

Here, the step of determining the estimated load index comprises a plurality of training data sets including a first training data set, wherein the first training data set comprises a sequence of kinematic information of a first entity during a first exercise session. preparing the artificial neural network to be trained based on a score comprising and labeled with a score collected from the first entity for an exercise load of the first exercise session; inputting the target data set to an input layer of the artificial neural network; and obtaining the estimated load index based on a result value of an output layer of the artificial neural network. can include

Here, the target-specific information includes an adjustment factor (modifier) for the estimated load index corresponding to the target entity, and the obtaining of the estimated load index includes adding the modifier to a resultant value of the output layer. It may be configured to obtain the estimated load index by multiplying by .

Here, the adjustment factor may reflect fitness evaluation information of the target entity.

Here, the adjustment factor is a game load index for the maximum value of the load index - the game load index is obtained using the artificial neural network based on kinematic information of the target entity during at least one recent game session. - can reflect the ratio of

Here, the adjustment factor may reflect evaluation information of the affiliated league of the target entity.

Here, the adjustment factor may reflect a ratio of the evaluation value of the affiliated league of the target entity to the evaluation value of the affiliated league of the first entity.

Here, the determining of the estimated load index may include a plurality of training data sets each including a first training data set and a first game data set - the first training data set is a first entity during a first exercise session and labeled with a score collected from the first entity for an athletic load of the first athletic session, wherein the first match data set comprises a sequence of kinematic information of the first entity during at least one competition session. preparing the artificial neural network trained based on a sequence of kinematic information and labeled with a maximum score; inputting the target data set and the target-specific information to an input layer of the artificial neural network; and obtaining the estimated load index based on a result value of an output layer of the artificial neural network. can include

Here, the artificial neural network may be trained using a fusion of a training feature map corresponding to the first training data set and a game feature map corresponding to the first game data set.

Here, the target-specific information may include a sequence of kinematic information during at least one recent game session of the target entity.

Here, the target-specific information includes identification information about a league to which the target entity belongs, and the determining of the estimated load index includes a plurality of training data corresponding to the league to which the target entity belongs, among a plurality of artificial neural networks. It can be configured to use an artificial neural network trained based on the set.

An apparatus for providing exercise load information on a target entity according to an aspect of the present disclosure includes a processor and a memory, and the processor includes a target-specific reflecting at least one of identification information or characteristic information on the target entity. obtain target-specific information; a target data set for the target entity during a target workout session comprising a plurality of time units, the target data set comprising: a sequence of kinematic information of the target entity for each of the plurality of time units; contains - obtains; and determining, from the target-specific information and the target data set, an estimated load index reflecting an exercise load degree of the target exercise session for the target entity, using an artificial neural network; can be configured to

According to one aspect of the present disclosure, in a non-transitory computer-readable storage medium storing instructions executable by a processor, the instructions are executed by the processor to cause the processor to: identify information about the target entity; obtain target-specific information reflecting at least one of the characteristic information; a target data set for the target entity during a target workout session comprising a plurality of time units, the target data set comprising: a sequence of kinematic information of the target entity for each of the plurality of time units; contains - obtains; and determining, from the target-specific information and the target data set, an estimated load index reflecting an exercise load degree of the target exercise session for the target entity, using an artificial neural network; can be configured to do so.

A method for providing exercise load information for a target entity according to an aspect of the present disclosure includes a target data set for a target entity during a target exercise session including a plurality of time units, wherein the target data set includes the plurality of time units. obtaining a sequence of kinematic information of the target entity for each of the units and condition information reflecting characteristics associated with the target exercise session; and determining an estimated load index reflecting an exercise load level of the target exercise session for the target entity, using an artificial neural network, from the target data set. can include

Here, the step of determining the estimated load index comprises a plurality of training data sets including a first training data set, the first training data set comprising: a sequence of kinematic information of a first entity during a first exercise session; and preparing the artificial neural network trained based on a score containing state information for the first exercise session and labeled with a score collected from the first entity for an exercise load of the first exercise session; inputting the target data set to an input layer of the artificial neural network; and obtaining the estimated load index based on a result value of an output layer of the artificial neural network. can include

Here, the state information may include daily readiness response information collected from the target entity for the target exercise session.

Here, the daily readiness response information includes a sleep time evaluation value; sleep quality evaluation values; fatigue level evaluation value; pain level (soreness level) evaluation value; stress level evaluation value; or a mood assessment value; may include at least one of them.

Here, the state information may include exercise type information reflecting information on whether the target exercise session is a match session or a training session.

Here, the status information may include an importance evaluation value according to the type of the target exercise session.

Here, the state information may include information about a length of time that has elapsed from a previous game session to the target exercise session.

Here, the status information may include information about a length of time that has elapsed from a start date of a sports season including the target exercise session to the target exercise session.

Here, the state information may include information about a physical condition of the target entity.

Here, the information on the physical condition includes age; key; weight; fitness rating; muscle strength evaluation value; maximum heart rate; or maximal oxygen uptake; may include at least one of them.

Here, the condition information may include environment information reflecting at least one of weather information, temperature information, humidity information, or season information for the target exercise session.

Here, the state information may include biological measurement information of the target entity during the target exercise session.

Here, the biological measurement information includes heart rate information; information about the ratio of heart rate to maximum heart rate; body temperature information; oxygen uptake information; or information about the ratio of oxygen uptake to maximal oxygen uptake; may include at least one of them.

An apparatus for providing exercise load information for a target entity according to an aspect of the present disclosure includes a processor and a memory, wherein the processor includes a target data set for the target entity during a target exercise session including a plurality of time units. - the target data set includes a sequence of kinematic information of the target entity for each of the plurality of time units and condition information reflecting a characteristic related to the target exercise session - obtain; and determining an estimated load index reflecting an exercise load level of the target exercise session for the target entity, using an artificial neural network, from the target data set. 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: A target data set for a target entity during a workout session, the target data set reflecting a sequence of kinematic information of the target entity for each of the plurality of time units and a characteristic associated with the target workout session. Obtaining - including condition information that and determining an estimated load index reflecting an exercise load level of the target exercise session for the target entity, using an artificial neural network, from the target data set. can be configured to do so.

background

In modern sports, a training method that improves performance by accumulating cycles of fatigue, recovery, and overcompensation through repetition of training and rest is widely used. It is important to properly adjust the exercise load of

Exercise load can be divided into external load and internal load. In general, the external load can be evaluated by considering the amount of exercise, such as total run-distance, and the intensity of exercise, such as the degree of performance of high-load exercise (eg, high speed running, heavy weight lifting, etc.). The load can be evaluated through the concept of a ratio of exercise effort performed against a sports player's maximum exercise capacity, such as heart rate reserve % (HRR%) or maximum oxygen intake (VO ₂ max).

While external load can be quantified relatively easily, internal load 1) requires a setup for measuring heart rate or oxygen intake, 2) it is difficult to express the measured values as a single indicator, and 3) the subject's condition or Due to the large error depending on the environmental conditions, it was difficult to quantify and objectify, so it was difficult to be used generally in the sports field despite the advantage of reflecting the characteristics of individual players.

Terms

Certain technical terms may be used in this description, and the following describes terms used in this description to establish definitional support for terms used in this description.

The following is a summary of preferred definitions of some terms used in this description. The following definitions are only exemplary definitions and are not exhaustive or limiting.

The term “sport player” refers to a person who performs sports activities, such as competitions, drills, or practices, and refers to the sport (e.g., soccer, basketball, football, baseball, boxing, track and field, swimming, etc.) and level of play (e.g., , professional, amateur, adult, youth, etc.) should be interpreted in a comprehensive sense to include all athletes (athletes) without limitation.

The term "attachable device" refers to an appliance that attaches directly or indirectly to a sports player. For example, the attachable device may be provided in the form of a pocket device inserted into a sports player's clothing or a wearable device in the form of a band wrapped around a sports player's body part, such as a wrist or ankle.

Exercise load information providing methods, devices, and systems according to embodiments of the present disclosure may provide exercise load information. Here, the exercise load information may mean information reflecting an exercise load induced by an exercise session, and specifically, but not limited to, the exercise load information is performed by the sports player(s) participating in the exercise session in the exercise session. It may refer to information reflecting an exercise load related to an activity. Exemplarily, exercise load information may be expressed in the form of a score or a single index, and methods, devices, and systems for providing exercise load information are based on information about the activity of a sports player who has performed an exercise session. A score or indicator can be provided that reflects the exercise load for the session.

system

1 illustrates a system for providing exercise load information according to an embodiment of the present disclosure. Hereinafter, an exercise load information providing system 1000 according to an embodiment of the present disclosure will be described with reference to FIG. 1 .

As shown in FIG. 1 , a system 1000 may include a sensing platform 1100 , a server 1500 and a terminal 1700 .

The system 1000 detects information about the sports player 10 participating in the exercise session during the exercise session through the sensing platform 1100, calculates exercise load information from the detected information through the server 1500, and the terminal ( 1700), it is possible to provide exercise load information by displaying the calculated exercise load information.

Hereinafter, components of an exercise load information providing system according to an embodiment of the present disclosure will be described.

sensing platform

The sensing platform 1100 can sense various information about the sports player 10 . For example, the sensing platform 1100 may perform localization of the sports player 10, detect a movement of the sports player 10, or sense a physical state of the sports player 10.

Illustratively, the sensing platform 1100 can sense kinematic information about the sports player 10 . The kinematic information is information about the position, posture, and movement of the sports player 10, and may include at least one of locational information, orientational information, and movement information. The motion information includes at least one of velocity, acceleration, jerk, angular velocity, angular acceleration, angular jerk, their magnitude (eg, speed in the case of velocity), and their direction (eg, in the case of velocity, the direction of movement). one may be included.

Illustratively, the sensing platform 1100 can sense physical condition information about the sports player 10 . The physical condition information is information reflecting the physical condition of the sports player 10, and may include at least one of heart rate, body temperature, blood pressure, respiratory rate, and electrocardiogram.

Hereinafter, information sensed by the sensing platform 1100 will be referred to as activity information.

The sensing platform 1100 may be provided in various forms to sense the above-described various types of information. Illustratively, the sensing platform 1100 may be a sensor-based platform using an attachable device 1300 attached to a sports player 10, an image-based platform using a camera 1200 disposed on or around a sports field, or the like. A combination of the two can be implemented.

Sensor-based platform

Hereinafter, a sensor-based sensing platform will be described.

A sensor-based sensing platform can include the attachable device 1300 . The sensing platform may obtain activity information about the sports player 10 to which the attachable device 1300 is attached by using a sensor mounted on the attachable device 1300 .

The attachable device 1300 can be attached to the sports player 10 and can be used to sense activity information of the sports player 10 . For example, the attachable device 1300 includes a Global Positioning System (GPS) sensor and can be used for positioning of the sports player 10 to which the attachable device 1300 is attached.

One or a plurality of attachable devices 1300 may be attached to one sports player 10 . In particular, when it is difficult to obtain all the information that the sensing platform wants to obtain through the attachable device 1300 with a single attachable device 1300, a plurality of attachable devices 1300 may be provided to one sports player 10. may need to be attached. For example, one attachable device including a GPS sensor and an Inertial Measurement Unit (IMU) sensor and another attachable device including a heart rate sensor are attached to the torso and wrist of the sports player 10, respectively. The attachable device attached to the torso can sense the position and movement of the sports player 10, and the attachable device attached to the wrist can sense the heartbeat.

Illustratively, the sensor-based sensing platform can obtain kinematic information using the attachable device 1300 .

Some examples of how a sensor-based sensing platform obtains kinematic information are described below.

The sensing platform may obtain location information using a positioning module of the attachable device 1300 .

For example, the attachable device 1300 includes a global positioning module (in other words, a satellite positioning module or a global navigation satellite system (GNSS) module), and the sensing platform performs global positioning using the global positioning module, so that the sports player 10 position can be measured. Specifically, the sensing platform is a sports player (10 ) can perform global positioning on . Meanwhile, the sensing platform may additionally include a base station for performing Real-Time Kinematic (RTK) for more accurate positioning. The sensing platform may use the global location as activity information, or may process the global location into a regional location defined as a reference coordinate system for the playground and use it as activity information. Here, the reference coordinate system may be a two-dimensional planar coordinate system having the length direction and the width direction of the playground as axes and the playground or a point around it (eg, one of the corners or the center of the stadium) as the origin. Meanwhile, it is disclosed in advance that all location-related information processed below can be processed according to a reference coordinate system without limitation.

In another example, the attachable device 1300 includes a local positioning module, and the sensing platform performs local positioning using a local positioning sensor network including the attachable device 1300 to determine the location of the sports player 10. position can be measured. The local positioning sensor network includes a tag node that is attached to an object that is a positioning target and moves, and an anchor node 30 that is fixedly installed in a positioning area, and is a Local Positioning System (LPS) that is transmitted and received between the tag node and the anchor node. Positioning can be performed using signals. The sensing platform transmits and receives LPS signals between an attachable device 1300 that includes a local positioning module and is attached to a sports player 10 as a positioning target and operates as a tag node and an anchor node 30 fixedly installed in or around a playground. Regional positioning of the sports player 10 can be performed using the result.

The sensing platform may acquire motion information and/or direction information using a motion sensing module of the attachable device 1300 .

For example, the attachable device 1300 includes an IMU sensor (or an Attitude and Heading Reference System (AHRS) sensor), and the sensing platform uses acceleration, angular velocity, and azimuth detected by the IMU sensor to be used by a sports player ( 10) or movement information and/or direction information of a body part of the sports player 10 may be acquired. When the attachable device 1300 is attached to the torso of the sports player 10, movement information according to the positional movement of the sports player 10 can be obtained, and when the attachable device 1300 is attached to the feet or legs of the sports player 10 Movement information of the arm or leg of the sports player 10 can be acquired.

Meanwhile, the sensing platform may calculate other kinematic information using the measured kinematic information. For example, the sensing platform may output speed based on the global position continuously measured by the GPS module and its sampling interval. Calculation of the kinematic information can be performed through simple mathematical operations including vector calculation or calculus on the time axis, so a detailed description thereof will be omitted. Here, the calculation of kinematic information can be performed internally in the attachable device 1300 or in the server 1500 receiving the information from the attachable device 1300, and a sensing module in the attachable device 1300. This can be performed directly or the controller of the attachable device 1300 can perform this based on the sensing result of the sensing module.

Illustratively, the sensor-based sensing platform can obtain body state information using the attachable device 1300 . In detail, the attachable device 1300 includes a biometric sensing module, and the sensing platform can detect a physical state of the sports player 10 using the biometric sensing module. Here, the biometric sensing module may include at least one of a heart rate sensor, a body temperature sensor, a blood pressure sensor, a respiration sensor, and an electrocardiogram sensor.

attachable device

2 is a block diagram of an attachable device according to an embodiment of the present disclosure.

As shown in FIG. 2 , an attachable device 1300 according to an embodiment of the present disclosure includes a sensing module 1310, a communication module 1320, a controller 1330, a memory 1340, and a battery 1350. can include

The sensing module 1310 can sense various signals to obtain activity information. The sensing module may be a positioning module, a motion sensing module, or a biometric sensing module.

The positioning module may be a satellite positioning module 1311 or a local positioning module 1313 . The satellite positioning module 1311 can perform positioning using a navigation satellite system, that is, GNSS. Here, GNSS may include Global Positioning System (GPS), GLONASS, BeiDou, Galileo, and the like. In detail, the global positioning module may include a satellite antenna for receiving a satellite signal and a positioning processor for performing positioning using the received satellite signal. For example, the satellite positioning module may be a GPS module including a GPS antenna for receiving GPS signals and a GPS processor for positioning using the GPS signals. In this case, the attachable device 1300 may operate as a GPS receiver, receive GPS signals from satellites, and obtain latitude, longitude, altitude, speed, azimuth, diluition of precision (DOF), and time from the GPS signals. there is.

The local location module 1313 can perform location in cooperation with a sensor network. The attachable device 1300 including a local positioning module can operate as a tag node of a local positioning sensor network and transmit/receive LPS signals with surrounding anchor nodes, and positioning can be performed using results of transmitting/receiving LPS signals. . For example, the attachable device 1300 transmits an LPS signal as a transmitter of a sensor network for local positioning and anchor nodes receive the LPS signal, or the anchor nodes transmit the LPS signal and the attachable device 1300 transmits the LPS signal to the local area. As a receiver of a sensor network for positioning, it can receive an LPS signal. In this case, the LPS signal may be periodically broadcast according to a communication method such as ultra-wide band (UWB) communication, Bluetooth, Wi-Fi, or RFID, and a time or device identifier may be implanted in the LPS signal. Position estimation may be performed based on measurement results according to transmission and reception of LPS signals, and RSS (Received Signal Strength) technique, ToA (Time-of-Arrival) technique, arrival time A time difference-of-arrival (TDoA) technique, an angle-of-arrival (AOA) technique, a triangulation technique, a hyperbolic triangulation technique, and the like may be used. Position calculation may be performed by a sensor network master, and any one of a tag node and an anchor node or a server may function as a sensor network master. If the sensing platform additionally includes an RTK base station, it is also possible that the base station is equipped with a position calculation function of the master.

The motion sensing module 1315 may also be referred to as a kinematic sensing module and can sense movement and/or posture. Here, the motion sensing module 1315 may include an IMU module or an AHRS module, where the IMU module and AHRS module include sensors including an accelerometer and a gyroscope and optionally a magnetometer. and the motion and posture can be measured from the detection result of the sensor.

The biometric sensing module 1317 can detect a body condition. Specifically, the biometric sensing module 1317 may include at least one of a heart rate sensor, a body temperature sensor, a blood pressure sensor, a respiration sensor, and an electrocardiogram sensor, and information on various body states may be measured using these sensors.

Meanwhile, not all of the above-described sensing modules must be provided in one attachable device 1300, and a plurality of sensing modules of the same type may be provided in one attachable device 1300, and the attachable device ( 1300) may have different sensing modules. For example, the sensing platform includes a GPS sensor and an Inertial Measurement Unit (IMU) sensor to acquire positional information and movement information about the sports player 10, and one attached to the torso of the sports player 10 and the sports player Two different attachable devices 1300 may be included, one containing a heart rate monitor to obtain a heart rate relative to (10) and the other attached to the wrist of the sports player (10).

The attachable device 1300 may transmit/receive data with an external device such as the server 1500 or another attachable device 1300 through the communication module 1320 . The communication module 1320 can perform wired communication and/or wireless communication. The attachable device 1300 is a wireless communication module that performs wireless communication of a mobile communication network (eg, LTE, 5G, etc.), Wi-Fi, Bluetooth, Zigbee, or other various standards. You can use it to send and receive data to and from external devices. For example, the wireless communication module 1320 transmits information acquired by the attachable device 1300 using the sensing module 1310 to a server (in real time) so that the system monitors the sports player 10 using the activity information. 1500), or when a plurality of attachable devices 1300 are attached to one sports player 10, data may be transmitted and received between the attachable devices 1300. In addition, the attachable device 1300 can transmit and receive data with an external device using a wired communication module that performs universal serial bus (USB) communication or wired local area network (LAN) communication. For example, the wired communication module may collectively transfer data collected by the attachable device 1300 to the server 1500 or a docking station that performs charging and/or data management of the attachable device 1300.

The controller 1330 can control overall operations of the attachable device 1300 . The controller 1330 may be implemented as a hardware configuration, a software configuration, or a combination thereof. From a hardware point of view, the controller 1330 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 1330 is not necessarily limited to a single physical entity, the controller 1330 is a processor that integrates and processes all processes of the attachable device 1300. Alternatively, it may be provided as a plurality of processors each performing different functions, or may be provided in a form combined with some of other components of the attachable device 1300. For example, the controller 1330 controls the operation and overall operation of a GPS processor that processes GPS signals to perform positioning, an IMU processor that performs various operations using results detected by sensors of the IMU module, and the attachable device 1300. It may be provided in a form including a main processor for controlling.

The memory 1340 can store various data related to operations of the attachable device 1300 . For example, firmware managing the operation of the attachable device 1300 or detection results of sensors of the attachable device 1300 may be stored in the memory 1340 . Memory 1340 can be provided as a variety of volatile and nonvolatile memories.

The battery 1350 can provide power required to drive the operation of the attachable device 1300 . The battery 1350 may be provided as a built-in or removable type, and may be charged by receiving power from an external power source.

Imaging-Based Sensing Platform

Hereinafter, an image-based sensing platform will be described.

An image-based sensing platform can include a camera 1200 . The image-based sensing platform may obtain various activity information through image processing and analysis of images captured through the camera 1200 .

The camera 1200 can be positioned on or around the playground to capture images of sports players 10 active on or within the playground. The camera 1200 may be installed semi-permanently (for example, installed in an accessory facility of a playground), temporarily installed (installed on a movable pole), or disposed in a form carried by a photographer. A sports video captured by the camera 1200 may be a tactical view, a broadcasting view, or a player-focused view. The tactical view is an image generally used for analyzing sports tactics, and may be an image captured so that most of the sports players 10 are included in the image for team tactical analysis, and the horizontal axis of the tactical view corresponds to the length direction or width direction of the stadium It can be. The broadcasting view is an image mainly used for sports relay and can generally be captured with a smaller angle of view than the tactical view. can be mainly used to analyze Also, the camera 1200 may not necessarily have a fixed viewing angle, and adjustment of the viewing direction and zoom may be performed manually or automatically.

An image-based sensing platform may include one or multiple cameras. A plurality of cameras may be provided in a multi-camera manner capable of capturing a panorama in a single spot, provided in a distributed arrangement in multiple spots, or provided in a combination of the two. .

The sensing platform may obtain activity information therefrom by analyzing an image captured through the camera 1200 . For example, the sensing platform performs object recognition for sports player 10 on the image, extracts the sports player 10's position within the image (eg, pixel coordinates), and camera 1200 Positional information on the sports player 10 can be obtained by projecting the position in the image onto the ground using the posture information of the . Here, a deep learning algorithm may be used for image processing such as object detection. Specifically, deep learning algorithms for object detection ranging from R-CNN (Region based Convolutional Neural Network) to YOLO (You-Only-Look-Once) are used for object detection (eg, sports player 10 detection) For coordinate conversion, for example, top-view conversion using camera parameters (eg, installation position, imaging posture, etc.) may be used. For example, the server 1500 acquires a sports image from a camera 1200 arranged to capture a tactical view, and uses an artificial neural network model that performs object detection to detect a ball or sports player 10 in the sports image. ), determine a representative pixel (eg, lower center of the bounding box) for the sports object in consideration of the bounding box, and consider camera parameters for the determined representative pixel in the sports image generated. Position data of the sports object may be obtained by performing top-view transformation using the coordinate transformation metric between the pixel coordinate system and the reference coordinate system. Here, object detection using an artificial neural network may be replaced with object segmentation. Accordingly, the server 1500 may obtain position data of a sports object in a sports image captured by the camera 1200 through image analysis.

In the above, the sensor-based sensing platform and the image-based sensing platform have been described, but as described above, in the present description, the sensing platform 1100 is sensor-based, image-based, or a combination thereof, as described above. can be implemented as In other words, the sensing platform 1100 may include at least one of the attachable device 1300 and the camera 1200 . When the sensing platform 1100 simultaneously operates the attachable device 1300 and the camera 1200, that is, when the sensing platform 1100 is implemented in a mixed form of a sensor-based platform and an image-based platform, Activity information may be obtained by simultaneously using information obtained through the attachable device 1300 and information obtained through image analysis. For example, the sensing platform 1100 may perform more precise positioning by comparing a position value obtained as a result of global positioning using the attachable device 1300 and a position value obtained through sports image analysis. will be.

Also, as mentioned in the above description, the sensing platform 1100 may optionally further include a base station for RTK correction and an anchor node for constructing a local positioning sensor network as needed.

Also, some of the various operations performed by the sensing platform 1100 may be processed by a processor or controller installed in the attachable device 1300, but other operations may be processed by the server 1500. For example, object detection for an image may be processed by the server 1500, and location calculation using a result of transmitting and receiving an LPS signal between nodes in a local positioning sensor network may be processed by the server 1500, In addition, various processing related to image analysis may be processed by the server 1500 . Accordingly, it may also be understood that the server 1500 is included in the sensing platform 1100 at this time.

server

The server 1500 may receive activity information from the sensing platform 1100 or obtain activity information in cooperation with the sensing platform 1100 and obtain exercise load information using the activity information.

The server 1500 may be provided as a local server located near a playground and/or a web server connected via the web. In addition, the server 1500 does not necessarily have to be implemented as a single subject. For example, the web server may be implemented by including a main server that processes calculations and a data server that stores various types of data. On the other hand, the local server may be provided as a device operated independently to perform only the original function of the server, but may be provided as a device having complex functions combined with other components of the exercise load information providing system. For example, the server may be provided in the form of a docking station, which is a container for accommodating, storing, and managing the attachable device 1300 . Here, the docking station may have a docking unit accommodating the attachable device 1300 and an internal space accommodating a plurality of attachable devices 1300, and when the attachable device 1300 is not in use, the attachable device (1300) can be stored. Furthermore, the docking station can provide various convenient functions required for use of the attachable device 1300 in addition to simply storing the attachable device 1300 . For example, the docking station can perform functions such as charging the attachable device 1300, displaying a battery status, updating firmware, and collecting data.

3 is a block diagram of a server according to an embodiment of the present disclosure.

The server 1500 can include a communication module 1510 , a controller 1520 and a memory 1530 .

The communication module 1510 can transmit/receive data between the server 1500 and other components or external devices of the exercise load information providing system. For example, the server 1500 collects data from the attachable device 1300 through the communication module 1510, receives an image from the camera 1200, or sends various information to the terminal 1700 through the web. can be conveyed

The controller 1520 can control the overall operation of the server 1500. Like the controller of the attachable device 1300, the controller 1520 of the server may be implemented as a hardware configuration, a software configuration, or a combination thereof, and from a hardware point of view, the controller may be an electronic circuit or an integrated circuit (IC). ), microchips, processors, etc. may be provided in various forms capable of performing calculations or data processing, and their physical configuration is not necessarily limited to a single physical entity. Meanwhile, specific operations performed by the server 1500 will be described later, and methods or operations according to an embodiment of the present disclosure described later can be understood as being performed by the controller 1520 of the server unless otherwise noted. make it clear in advance that

However, as seen in the present description, the embodiments according to the present disclosure are not necessarily limited to being performed by the controller 1520 of the server, and the embodiments according to the present disclosure are, for example, the controller of the attachable device 1300 It should be understood that the processing may be performed by a processor capable of arithmetic operations, such as in step 1330, or at least some procedures may be performed by a server and at least some other procedures may be performed by other processors.

terminal

The terminal (or terminal device) 1700 can function as a user interface that provides various data or information collected or calculated by the system to the user or receives user input from the user. For example, the terminal 1700 receives a user input indicating a specific exercise session for which exercise load information is desired from the user, requests exercise load information on the corresponding exercise session from the server 1500, receives a reply, and displays it. By doing so, it is possible to provide exercise load information to the user. The terminal 1700 can be a smart device such as a smart phone or tablet, a personal computer such as a laptop or desktop, or other electronic device having an input interface for receiving user input and an output interface such as a display.

How to provide exercise load information

Hereinafter, a method for providing exercise load information according to an embodiment of the present disclosure will be described. In the following description, although the exercise load information providing method is described as being performed by the above-described exercise load information providing system, this is only for convenience of explanation, and thus the method is not limited by the system.

exercise load information

In the present description, exercise load information refers to information reflecting the exercise load of a specific exercise session, and may be exemplarily expressed as an exercise load index representing an exercise or a degree of exercise load as a score. For a non-limiting example, the exercise load information may refer to information reflecting exercise load related to activities performed by sports player(s) participating in the exercise session in the corresponding exercise session.

Exercise load can be divided into external load, which expresses the physically performed exercise as an objective physical quantity, and internal load, which considers the degree of effort experienced subjectively by the subject who performed the exercise. A representative example of the internal load is the perceived exercise ( and RPE: Rating of Perceived Exertion (hereinafter, also referred to as 'perceived fatigue'). In order to accurately measure the internal load, it is necessary to consider how hard the body of the exercise subject feels the exercise based on the physical sensation and feeling of the subject who performed the exercise. In order to quantify the degree to which the body feels about exercise, that is, the internal load, the degree of increase in heart rate, the degree of increase in breathing rate, the degree of temperature increase, the degree of accumulation of muscle fatigue, the degree of pain, etc. are measured, and some or all of these are measured individually or collectively. Attempts have been made to quantify However, since the internal load is affected by very complex factors that act on the human body, it is difficult to accurately calculate only with the above-mentioned information, and it is practically impossible to measure the above-mentioned information from the exercise subject for each exercise session. is a method of grasping the internal load for exercise through a questionnaire about the intensity felt by the exercise subject for the exercise session after the end of the exercise session, that is, the extent to which the exerciser perceived the exercise intensity is generally used.

In this description, the exercise load information provided by the system includes information indicating the degree of external exercise load, information indicating the degree of internal exercise load, or information indicating the degree of comprehensive exercise load of internal / external loads. It should be understood as a comprehensive concept. As will be described later, in this system, the exercise load information may be seen as a concept close to reflecting the internal load in terms of being calculated by considering the exercise awareness experienced by the subject who performed the exercise, but the exercise awareness obtained from a number of exercise subjects Objectivity is ensured in terms of calculation by using the comprehensively, and external loads are reflected when considering the aspects calculated based on physical quantities such as speed or acceleration according to the movement performed by the exercise subject during the exercise session. may be Therefore, in the present description, the exercise load information may be interpreted as information reflecting the internal load for exercise, but depending on the point of view, it may be interpreted as information reflecting the external load independently or additionally, so the exercise load information of the present description It should be noted in advance that the reflected exercise load should not be limitedly interpreted as an internal load and/or an external load.

4 illustrates an example of an exercise load index according to an embodiment of the present disclosure.

The exercise load index may be expressed as an exercise load index representing the degree of exercise load as a score. Here, the expression form of the exercise load index may be the same as or similar to RPE. Specifically, the exercise load index may be an index expressed as a specific score within a predetermined score range indicating the intensity of exercise or the magnitude of exercise load for each score. For example, the exercise load index may be expressed as an RPE of a BORG scale or a BORG-CR10 scale. Here, the RPE of the Bolg scale is a scale that expresses the state of no exercise load to the state of extreme exercise load from 6 to 20 points, and the Bolg-CR10 scales from no load to maximum load between 1 and 10 points. scale to express Meanwhile, when the RPE value of the Bolg scale is multiplied by 10, the heart rate of the exercising subject according to the exercise expressed in the Bolg scale can be roughly predicted. In the present description, the exercise load index provided by the system may express the exercise load through a score of a predetermined range that is the same as or similar to the RPE of the Bolg scale or the Bolg-CR10 scale described above. However, in this description, the exercise load index does not necessarily have to be expressed in this form, and the percentile for a specific exercise characteristic (eg, the percentile for the exercise effort performed against the maximum physical strength or maximum exercise effort capacity of a sports player), physical quantity It can be expressed in various index forms, including (for example, calories consumed by exercise) or letter grades such as grades.

In the present disclosure, the exercise load index may include an estimated load index or a response load index. The “estimated load index” may be an exercise load index determined using an artificial neural network from a data set according to embodiments of the present disclosure. The “response load index” may be an exercise load index received in response from a target entity to be measured after the end of a target session for which exercise load is to be measured.

Hereinafter, embodiments of a method for providing exercise load information according to the present description will be described in more detail. 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.

Exercise load determination based on kinematic information

5 is a schematic flowchart of a method for providing exercise load information according to an embodiment of the present disclosure. 6 is a detailed flowchart of the step of determining the estimated load index in FIG. 5; Hereinafter, a method for providing exercise load information according to an embodiment of the present disclosure will be described in more detail with reference to FIGS. 5 and 6 .

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 server 1500 as described with reference to FIGS. 1 to 3 or the controller 1520 included in the server 1500, and includes a processor and a memory. Any computing capable device may be used, and a combination of a plurality of physically separate devices may be referred to as a computing device.

According to the exercise load information providing method according to an embodiment of the present disclosure, an estimated load index may be determined based on kinematic information of a target entity during a target exercise session for which load information is to be provided and provided as exercise load information. there is. Accordingly, information on the degree of exercise load of the target entity during the target exercise session can be conveniently collected and managed without directly querying the target entity.

In the present description, “target entity” may mean an object for which exercise load information is to be determined. For example, the target entity may be a single sports player, and information about how much exercise load the individual player felt during a workout session may be provided. Alternatively, the target entity may be a sports team including a plurality of sports players, and information about how much exercise load the entire corresponding team experienced during a workout session may be provided.

Also, in the present description, “target exercise session” may refer to a target exercise session for which exercise load information is to be provided. An exercise session may be a section having a predetermined length of time and may include a plurality of time units. At least some of the basic data that is the target for determining the exercise load information may be determined every such plurality of time units. Accordingly, at least some of the basic data may include a sequence of predetermined information. A set of basic data that is a target for determining exercise load information may be referred to as a “target data set” in the present disclosure.

Referring to FIG. 5 , a method for providing exercise load information according to an embodiment of the present disclosure includes measuring location information of a target entity (S510), obtaining a target data set (S520), and determining an estimated load index. It may include one or more of the step of determining a representative load index (S530) or the step of determining a representative load index (S540).

Hereinafter, each step of an embodiment of a method for providing exercise load information will be described.

As shown in FIG. 5 , the computing device may measure location information of a target entity (S510). For example, the computing device may measure location information of a target entity for a target exercise session using a positioning device corresponding to the target entity. Here, the positioning device may mean at least a part of the sensing platform 1100 as exemplarily described above in this description. Accordingly, the computing device may obtain location information of the target entity at predetermined time intervals during the target exercise session using, for example, a sensing platform using one or more of the attachable device 1300 and/or the camera 1200 attached to the sports player. can be measured Thus, a sequence of location information of a target entity during a target workout session may be obtained. The sequence of location information may include, for example, location information of a target entity at a first time point, location information of a target entity at a second time point, and location information of a target entity at an n-th time point.

Referring back to FIG. 5 , the computing device may acquire a target data set for the target exercise session (S520). More specifically, the computing device may obtain a target data set for a target entity during a target workout session comprising a plurality of time units, wherein the target data set includes a target entity for each of the plurality of time units. may include a sequence of kinematic information of For example, kinematic information may be generated based on location information about a target entity.

According to one aspect, a target exercise session may include a plurality of time units having a predetermined time interval, and kinematic information of a target entity may be determined for each time unit. Here, location information for one viewpoint does not have to correspond to each time unit for which kinematic information is determined. For example, kinematic information of a target entity for a first time unit may be determined based on location information of a target entity of multiple views corresponding to a first time unit, and kinematic information of a target entity of multiple views corresponding to a second time unit may be determined. Kinematic information of the target entity for the second time unit may be determined based on the location information of the target entity.

As discussed above in this description, the kinematic information may represent information on at least one of the location, posture, and motion of the target entity, and may include at least one of location information, direction information, and motion information. Motion information will be detailed later in this description.

The sequence of kinematic information included in the target data set may include, for example, kinematic information of the target entity for the first time unit, kinematic information of the target entity for the second time unit, and kinematic information of the target entity for the n-th time unit. May include kinematic information. While the conventional measurement of external load is based on statistical values for the entire exercise session, such as the total amount of movement, exercise time, and average speed during an exercise session, the provision of exercise load information according to an embodiment of the present disclosure is based on a target exercise session. It can be distinguished in that an evaluation value for the degree of exercise load is determined based on a sequence of kinematic information for a plurality of included time points and / or time units.

Referring back to FIG. 5 , the computing device may determine an estimated load index (S530). More specifically, the computing device may determine, from the target data set, using an artificial neural network, an estimated load index that reflects the degree of exercise load of the target exercise session for the target entity. According to one aspect, the determined estimated load index may be provided as an estimate of exercise load for a target exercise session. Here, the estimated load index may include, for example, a plurality of numerical values indicating the degree of exercise load, and may be configured such that 10 represents the highest exercise intensity with a value of 1 to 10, for example, but in this specific numerical range Not limited.

According to an embodiment of the present disclosure, by automatically determining an estimated load index using an artificial neural network based on target data sets measured and/or calculated for the target entity, manual query and response for the target entity Information on the degree of exercise load of the target entity during the target exercise session may be conveniently obtained and provided without going through a procedure.

An estimated load index according to an aspect of the present disclosure may be an estimate of a rate of perceived fatigue (RPE) of a target entity for a target exercise session. For example, an estimated load index determined according to an aspect of the present disclosure may be provided instead of the perceived fatigue as exemplarily illustrated in FIG. 4 .

Relatedly, FIG. 6 is a detailed flowchart of the step of determining the estimated load index in FIG. 5 . As shown in FIG. 6, the computing device first prepares a trained artificial neural network (S531), inputs a target data set to the input layer of the artificial neural network (S533), and based on the resultant value of the output layer of the artificial neural network Thus, an estimated load index may be obtained (S535). Hereinafter, with reference to FIG. 6, the estimated load index determination process will be described in more detail.

As shown in FIG. 6 , the computing device may prepare a trained artificial neural network based on a plurality of training data sets including the first training data set (S531). Specifically, for example, an artificial neural network model may be stored or loaded in a memory of a server, and the controller may be in a state capable of operating the artificial neural network model.

7 illustrates an exemplary artificial neural network used for calculation of an exercise load index according to an embodiment of the present disclosure.

An artificial neural network may include an input layer, a hidden layer, and an output layer. An input layer may include a plurality of nodes and may receive an input data set. The output layer may include at least one node, and, for example, has a form having a single node that outputs an output value that can be interpreted as a value of an exercise load index as a numerical value, or different values of the exercise load index are assigned to each. and can be provided in the form of having multiple nodes whose output values are probability values. The hidden layer may include a plurality of layers connecting the input layer and the output layer and having at least one node. Connections between nodes may be defined by parameters such as weights or biases, and through this, an output value may be calculated from an input data set while information is transferred from an input layer to an output layer via a hidden layer.

Learning of the artificial neural network may be performed using a training data set and an exercise load index labeled therein. Specifically, learning adjusts the parameters of the artificial neural network in the direction of reducing the error based on the error between the result value output by the artificial neural network receiving the training data set and the ground truth of the exercise load index labeled in the training data set It can be done through

In one embodiment of the present disclosure, the artificial neural network may be trained using a training data set labeled with an exercise load index to output an exercise load index of a target exercise session when activity information of a target entity for a target exercise session is input. can Basically, the learning data set may be generated from activity information about a target entity that has performed an exercise session, and an exercise load index to be labeled may be a value collected from the target entity that has performed an exercise session.

More specifically, but not limitedly, the artificial neural network according to an embodiment of the present disclosure may be trained based on a plurality of training data sets including the first training data set. Here, the first training data set may include a sequence of kinematic information of the first entity during the first exercise session, and may be labeled with a score collected from the first entity for an exercise load of the first exercise session. .

In relation to this, FIG. 8 illustrates an example of a training data set of an artificial neural network according to an embodiment of the present disclosure, and FIG. 9 is a detailed view of an example of a kinematic information sequence of FIG. 8 .

As shown in FIG. 8 , a plurality of training data sets 800 for training an artificial neural network according to an aspect of the present disclosure may include, for example, a first training data set 810 . The first training data set 810 includes a sequence of kinematic information 811 of a first entity for each of a plurality of time units included in the first workout session, from the first entity during the first workout session. It can be labeled with a score 813 collected for exercise load. That is, for example, when sports player A participates in a specific workout session A, a sequence of kinematic information for each of the time units for sports player A may be collected and/or measured. Here, after the end of the corresponding exercise session A, for example, as a measurement of perceived fatigue (RPE), an exercise load evaluation value may be inquired from sports player A and a response may be received. Here, one training data set can be secured by labeling the sequence of kinematic information of sports player A with the received RPE value (eg, 8).

As shown in FIG. 9 , the sequence 811 of kinematic information of the first entity is a plurality of time units included in the first workout session, eg, the first time unit 811a to the nth time unit. kinematic information (eg, velocity, acceleration, jerk, angular velocity, angular acceleration, angular jerk) of the first entity measured for each of the time units leading to 811n.

By learning an artificial neural network based on a plurality of training data sets as shown in FIGS. 8 and 9 , when a target data set is input, a corresponding exercise load evaluation value can be output.

More specifically, referring back to FIG. 6 , the computing device may input the target data set to the input layer of the artificial neural network (S533). As described above, the target data set may include a sequence of kinematic information of a target entity for a plurality of time units during a target exercise session.

Meanwhile, according to one aspect of the present disclosure, zero padding may be performed on an input data set (eg, a target data set) so that it corresponds to the number of nodes of an input layer of an artificial neural network. For example, a target data set may be configured to have a predetermined length by performing zero-padding before or after a sequence of kinematic information. In relation to this, FIG. 10 illustrates an example of an input data set of an artificial neural network according to an embodiment of the present disclosure, and FIG. 11 illustrates another example of an input data set of an artificial neural network according to an embodiment of the present disclosure. . FIG. 10 shows that zero padding is performed before a sequence, and FIG. 11 shows that zero padding is performed after a sequence.

In other words, the artificial neural network according to the embodiments of the present disclosure may be configured to receive data of a fixed shape or length as input, and zero-padding may be performed on the feature sequence to fit the input data set to the determined length. there is. In this regard, considering that a typical game or training lasts for about 2 hours, the maximum length is set to, for example, 2,000, and for a target data set having a length shorter than this, to fill the front or back of the kinematic information sequence Multiple '0's can be added. If '0' is padded at the end of the sequence, it may correspond to a situation in which the target entity recognizes the RPE score after resting for the zero-padded period. That is, the shorter the original sequence of the model, the longer the break between the end of the activity and the examination of the RPE is likely to be misrepresented. Thus, it may be configured to perform zero-padding at the front of the sequence rather than at the back, more preferably in the present description.

An artificial neural network outputs a result value to an output layer in response to receiving a target data set. The computing device may acquire an estimated load index based on a result value of the output layer of the artificial neural network (S535). According to one aspect, an estimated load index may be provided as information indicating a degree of exercise load of a target entity according to a target exercise session.

In this way, by determining an estimated load index using an artificial neural network based on a target data set collected for a target entity and providing it as exercise load information, it is possible to conveniently replace a conventional manual RPE collection process. Therefore, it is possible to improve RPE errors (eg, fluctuations in RPE collection time after training or occurrence of players who have not submitted) due to collection according to the existing question-and-answer method, and predetermined, such as daily, weekly, monthly units. It is possible to manage the exercise load information for a period having a continuity of. Therefore, it is possible to set up a training session more efficiently, and an improvement in performance can be expected. In the prior art, it was possible to manage exercise load information in units of teams substantially, but according to the embodiment of the present disclosure, it is possible to manage exercise load at the level of individual players, so that individual players' conditioning can be performed.

On the other hand, according to one aspect of the present disclosure, it is also possible to obtain and provide exercise load information for a sports team. According to one aspect, a target entity for determining the estimated load index may be at least one sports team.

In this case, referring to FIG. 5 again, the computing device may determine a representative load index (S540). More specifically, the computing device may determine a representative load index during a target exercise session for the sports team based on the estimated load indexes for each of at least some athletes belonging to the sports team (S540). For example, if the target entity is sports team A, the steps of measuring location information for sports players A, B, and C belonging to the target entity (S510), obtaining a target data set (S520), and determining an estimated load index (S530) It is possible to determine each estimated load index through Thereafter, a representative load index for sports team A may be determined using estimated load indexes of each of sports players A, B, and C. The representative load index may be, for example, an average of estimated load indexes of each sports player, but is not limited thereto, and a representative value indicating a trend of the entire team may be determined according to various rules.

As described above, according to an embodiment of the present disclosure, an estimated load index reflecting an estimated value of an exercise load degree for a target entity may be determined based on, for example, kinematic information of the target entity.

As discussed above in the present disclosure, exercise load information such as an estimated load index may be, for example, an estimated value for Rated Perceived Exertion (RPE). Perceived fatigue may reflect the degree of difficulty in exercise. Accordingly, at least one of a variety of kinematic information may be used to determine exercise load information, such as, for example, an estimated load index.

For example, in determining the level of difficulty of the exercise, at least one of i) high energy consumption, ii) high force, and iii) instantaneously rapid change may be considered.

Regarding the dissipation of energy, kinetic energy is defined as:

K = 1/2 mv ²

That is, for example, as an evaluation factor for energy consumption, the velocity v of the target entity may be included in the kinematic information.

Regarding the force required, Newton's second law is defined as:

F = ma

That is, for example, the acceleration a of the target entity may be included in the kinematic information as an evaluation factor for required force.

Meanwhile, muscle fatigue may be mainly affected by jerk. Therefore, the jerk of the target entity may be included in the kinematic information as an evaluation factor of fatigue due to an instantaneous rapid change being applied.

That is, the kinematic information may include, for example, at least one of velocity, acceleration, or jerk of the target entity.

Unlike conventional methods for determining external loads that have been evaluated through fragmentary values such as total moving distance or average speed according to statistical results, the method for providing exercise load information according to an embodiment of the present disclosure is a method for determining exercise load information. By inputting a sequence of kinematic information (for example, at least one of speed, acceleration, or jerk) for each of a plurality of time units included in the target exercise session to the artificial neural network, the information about the exercise load with more improved accuracy evaluation values can be estimated.

Meanwhile, even for the same magnitude of kinematic elements (for example, the same displacement, magnitude of velocity, magnitude of acceleration, or magnitude of jerk), the motion load applied to the target entity may be different according to the direction of the kinematic element. . For example, even when moving the same distance at the same speed, the target entity moves toward the front of the target entity, moves backward in the direction opposite to the orientation direction of the target entity, or moves laterally relative to the orientation direction of the target entity. Moving towards may have different exercise loads applied to the target entities. For example, moving backward in a direction opposite to the target entity orientation direction may cause the target entity to experience a greater motion load than moving forward in the orientation direction at the same displacement or velocity. Accordingly, according to one aspect of the present disclosure, a relation of a direction of a kinematic element to an orientation direction of a target entity may be considered in determining exercise load information.

In this regard, kinematic information according to an embodiment of the present disclosure may be expressed based on any one of a plurality of coordinate systems. For example, kinematic information according to one aspect may be expressed based on a field-based coordinate system, or may be expressed based on an entity-based coordinate system, It is not limited to this.

Relatedly, FIG. 12 illustratively illustrates an entity-based coordinate system. As shown in FIG. 12 , the entity-based coordinate system may be a coordinate system set based on an orientation direction of a target entity such as a sports player, for example. For example, an entity-based coordinate system may include a forward and backward axis 121 corresponding to a traveling direction of a target entity and a left and right axis 123 corresponding to a lateral direction of the target entity. When the target entity moves forward, for example, the value for the displacement may have a positive value in the forward and backward axis 121 direction when expressed based on the entity-based coordinate system, and the target entity moves in a direction opposite to the orientation direction. When moving backward, a displacement value for this may have a negative value in the direction of the front and rear axis 121 . Meanwhile, the displacement generated to the left or right based on the orientation direction of the entity may have a negative value or a positive value in the direction of the left and right axes 123, respectively. Similarly to the velocity, acceleration, or jerk that can be calculated based on displacement, the movement direction relative to the orientation direction can be expressed only with the value of kinematic information.

Referring to FIG. 12 , for example, a first movement (m1) and a second movement (m2) are shown. The first movement may be a forward movement in the entity orientation direction, and the second movement may be a leftward side step movement based on the entity orientation direction. Here, the first motion may be expressed as a displacement having a positive value in the forward and backward axis 121 direction, and for example, a motion vector may be expressed as (0, 10). In addition, the second motion may be expressed as a displacement having a positive value in the left and right axis 123 direction, and for example, a motion vector may be expressed as (10, 0).

That is, when the kinematic information expressed based on the entity-based coordinate system is information having direction and magnitude, such as displacement, velocity, acceleration, or jerk, information on the direction of movement relative to the orientation direction of the entity can include Accordingly, kinematic information according to an aspect of the present disclosure may include first kinematic information related to the velocity of the target entity, second kinematic information related to the acceleration of the target entity, or third kinematic information related to the jerk of the target entity. At least one of kinematic information may be included. Even if the kinematic information includes only information about linear movement such as velocity, acceleration, or jerk, if the information about such linear movement is expressed based on the entity-based coordinate system, the movement in the orientation direction of the entity A difference in the degree of exercise load according to the direction of (eg, forward, backward, or lateral movement) may be reflected.

In order for kinematic information to be expressed based on an entity-based coordinate system, information about an orientation direction of an entity is required. As described above in the present description, the sensing platform according to one aspect may include, for example, an attachable device, and the attachable device may include, for example, an inertial sensor such as an inertial measurement unit (IMU) sensor. . According to one aspect of the present disclosure, kinematic information may be obtained based on information measured by an inertial sensor corresponding to a target entity and/or information measured by a location measuring device such as a GPS sensor.

On the other hand, according to one aspect of the present disclosure, kinematic information may be obtained using location information of a target entity measured using a location measuring device such as a GPS sensor. For example, in consideration of economic aspects or various circumstances such as at least one of improvement in wearing comfort, light weight, and reduction in computation, the sensing platform may not include an inertial sensor and may include only a position measurement device such as GPS. In this case, since information on the orientation of the target entity cannot be obtained, kinematic information may be expressed based on a stadium-based coordinate system rather than an entity-based coordinate system.

Here, the stadium-based coordinate system may represent a local location as described above in this description. For example, the local location may be obtained by converting coordinates according to a global positioning device such as GPS through operation, or directly obtained through a local positioning module, but is not limited thereto.

More specifically, FIG. 13 illustrates an arena-based coordinate system by way of example. As shown in FIG. 13 , a stadium-based coordinate system can include a first axis 133 for the length of the field on which the athletic session is conducted and a second axis 131 for the width of the field for the athletic session to be conducted. .

The kinematic information for linear motion expressed according to the arena-based coordinate system does not reflect the relationship between the direction of orientation and the direction of movement of the target entity. For example, as shown in FIG. 12 , the first movement (m1) may be a forward movement in the orientation direction of the entity, and the second movement (m2) may be a side step movement toward the left with respect to the orientation direction of the entity. However, as shown in FIG. 13, when the same first motion (m1) and second motion (m2) are expressed through the stadium-based coordinate system, they are only expressed as motions toward different predetermined directions, respectively, for the orientation direction of the entity. Information about the direction of motion (eg, forward, backward, or lateral motion) cannot be expressed. Therefore, for example, although the exercise loads for the forward motion and the side step may actually be different, they may be incorrectly reflected as having the same exercise load in determining the estimated load index.

In this regard, according to one aspect of the present disclosure, an estimated load index may be determined by reflecting a relationship between an alignment direction and a moving direction of a target entity based on a difference between a moving direction of a previous view and a moving direction of a current view. For example, when the target exercise session includes a plurality of time units, in addition to information about the linear motion of the current time unit, the direction of information about the linear motion of the previous time unit and information about the linear motion of the current time unit An estimated load index may be determined by further reflecting the difference from the direction of . In other words, the kinematic information for a second time unit, for example, is information about the linear motion of the target entity in the second time unit (eg, information about velocity, acceleration, jerk, or magnitude thereof). and information on a difference between at least one direction of the information on the linear motion of the target entity in the first time unit and at least one direction of the information on the linear motion of the target entity in the second time unit. can do. Accordingly, for example, even when kinematic information is expressed through a stadium-based coordinate system, an estimated load index may be determined by reflecting a relationship between an alignment direction and a motion direction of a target entity.

14 illustrates a difference between a motion direction of a previous view and a motion direction of a current view. Looking more specifically with reference to FIG. 14 , it can be assumed that the movement direction 141 of the previous view is the alignment direction of the target entity of the current view. Since the movement direction of the target entity during the exercise session is most often observed forward, if it is assumed that the target entity moves forward at the previous time point, the direction of the movement at the previous time point (e.g., the speed at the previous time point) direction) can be assumed to be the same as the orientation direction of the current viewpoint of the target entity. Here, considering the motion direction 143 of the current view as illustrated in FIG. 14 , the motion direction of the current view (eg, the direction of speed of the current view) 143 and the direction of motion of the previous view 141 The angular change amount 145 of the liver can be used as a factor reflecting a relationship between the alignment direction of the target entity at the current view and the movement direction of the target entity at the current view.

Accordingly, according to an embodiment of the present disclosure, kinematic information for a target time unit, which is any one of a plurality of time units included in the target exercise session, is a size of at least one of linear motions of a target entity in the target time unit. It may include information about , and information on an angular change amount between at least one direction of the linear motion of the target entity in the target time unit and at least one direction of the linear motion of the target entity in the previous time unit of the target time unit. there is. For example, as exemplarily shown in FIG. 14 , the kinematic information for the target time unit includes information about the magnitude (speed) 147 of the velocity of the target time unit and the previous time unit of the target time unit. It may include information about the angular change amount 145 between the direction of velocity 145 in the target time unit and the direction 143 of velocity in the target time unit.

According to one aspect, the kinematic information for the target time unit includes speed for the target time unit, change in direction of movement from the previous time unit, magnitude of acceleration, change in direction of acceleration from the previous time unit, magnitude of jerk, It may be configured to include at least one or more of changes in the direction of the jerk from the previous time unit. 15 shows an example of kinematic information according to an embodiment of the present disclosure. As shown in FIG. 15 , kinematic information can be obtained by calculating based on positional information of a target entity measured based on the positioning sensor 15 . The acquired kinematic information includes the horizontal speed of the target entity (151), change in horizontal movement direction (152), magnitude of horizontal acceleration (153), change in horizontal acceleration direction (154), horizontal jerk It may include at least one of magnitude 155 or change in horizontal jerk direction 156 . Each kinematic information may be obtained for each of a plurality of time units included in the target exercise session to form a sequence 811 of kinematic information included in the target data set.

Hereinafter, more various examples of kinematic information according to embodiments of the present disclosure will be described.

16 shows an example of kinematic information according to an embodiment of the present disclosure. As shown in FIG. 16, kinematic information 1600 includes position/orientation information 1610, linear movement information 1620, and angular movement information 1630. may include at least one of them.

In relation to this, FIG. 17 shows examples of coordinate systems capable of expressing kinematic information according to an embodiment of the present disclosure. As shown in FIG. 17, kinematic information according to one aspect may be expressed by any one of, for example, a global coordinate system 171, a local coordinate system 173, an inertial coordinate system 175, and a hybrid coordinate system 177. there is.

Global coordinate system 171 can include latitude, longitude, and altitude components measured using a global positioning module such as, for example, GPS. Local coordinate system 173 can include, for example, the arena-based coordinate system described above in this description. The local coordinate system may include an axis in the longitudinal direction of the stadium in which the exercise session is performed and an axis in the width direction of the stadium. According to one aspect, in addition to this, a vertical axis from the ground of the stadium may be included. Further, according to one aspect, a specific point in the stadium, such as the center or edge of the stadium, may be used as the origin.

The inertial coordinate system 175 can be, for example, an aviation coordinate system. The inertial coordinate system may include forward-backward movement, slide movement, and up-down movement elements of the player's body. In addition, it may include rotational movement elements of roll, pitch, and yaw axes related to rotation of the player body.

Meanwhile, the hybrid coordinate system 177 may be, for example, a maritime coordinate system. The hybrid coordinate system is fixed in the direction of the yaw axis and may include heading in the horizontal direction, lateral movement in the horizontal direction, and elements in the vertical direction.

A coordinate system for expressing kinematic information according to an aspect of the present disclosure is not limited to the above-described coordinate systems, and kinematic information may be expressed based on any one of various coordinate systems.

Referring again to FIG. 16 , kinematic information 1600 can include position/orientation information 1610 . Information about position/orientation may not, per se, affect exercise load. However, the position/orientation information can be data for calculating various kinematic information. Information about position/orientation may also be referred to as static information. 18 shows an example of static information. Static information 180 may include position information 181 . For example, location information 181 may include latitude, longitude, and altitude, or may include length, width, and height. Meanwhile, the location information 181 may include orientation information 183 . Orientation information 183 can include, for example, longitudinal orientation, width orientation, and height orientation. Such static information may be obtained for each of a plurality of time points or time units included in the target session to form a sequence. Additional kinematic information (eg, linear motion information or angular motion information) can be obtained based on the sequence of static information.

In relation to this, referring to FIG. 16 , kinematic information 1600 may include information 1620 related to linear movement. As discussed above in this description, the information on the linear movement may be expressed based on, for example, a stadium-based coordinate system, an entity-based coordinate system, or an arbitrary coordinate system.

More specifically, the information about linear motion can include first kinematic information 1621 related to the velocity of the target entity. According to one aspect of the present disclosure, the first kinematic information is information about the magnitude of velocity in a target time unit, or information between a direction of velocity in a previous time unit of the target time unit and a direction of velocity in the target time unit. At least one of information about the amount of angular change may be included.

19 shows an example of first kinematic information including speed-related information.

As exemplarily described above in the present description, when kinematic information is expressed based on an entity-based coordinate system, even if an estimated load index is determined using only speed information, the motion load according to the difference between the orientation and the movement direction of the entity Differences in degree may be reflected. Accordingly, the first kinematic information may include velocity information (v1, v2, v3) of the target entity. The speed information may be expressed based on various coordinate systems, for example, speed information based on longitudinal, transverse, and height coordinates (v1), orientation, lateral, and vertical direction coordinates (v2), At least one of forward, backward, leftward, rightward, upward, and downward coordinate-based speed information (v3) may be included in the first kinematic information as the speed information.

On the other hand, when kinematic information is expressed based on, for example, a field-based coordinate system, a difference between an orientation and a movement direction of an entity cannot be reflected in determining an exercise load using only speed information. In this case, the first kinematic information may include speed magnitude (velocity) information (s1, s2, s3, s4) and movement direction change information (vd1, vd2). The speed information may be expressed based on various coordinate systems, for example, the magnitude of the speed based on orientation direction, lateral direction, and vertical direction coordinates, that is, speed information (s1), forward, backward, lateral direction, and vertical direction coordinates. Speed information (s2), horizontal direction speed information (s3), which is magnitude information about the sum speed of the longitudinal speed and width direction speed, and total speed, which is magnitude information about the sum speed of the longitudinal speed, width direction speed, and height direction speed At least one of the information (s4) may be included in the first kinematic information as speed information. In addition, motion direction change information may also be expressed based on various coordinate systems, and may be determined from, for example, previously sampled orientations. For example, a horizontal movement direction (vd1), which is an angular difference between a velocity direction of a previous time unit and a velocity direction of a current time unit, may be utilized as movement direction change information. Alternatively, the magnitude (vd2) of the angular difference between the velocity direction of the previous time unit and the velocity direction of the current time unit may be utilized as motion direction change information.

According to one aspect, the information about the linear motion can include second kinematic information 1623 related to the acceleration of the target entity. According to one aspect of the present disclosure, the second kinematic information is information on the magnitude of acceleration in the target time unit, or between the direction of acceleration in the previous time unit of the target time unit and the direction of acceleration in the target time unit. At least one of information about the amount of angular change may be included.

On the other hand, in the information about the acceleration, the magnitude of the acceleration in the target time unit is used, but in the information for reflecting the relationship between the orientation direction and the direction of movement of the target entity, the change in the velocity direction rather than the change in the acceleration direction may be more suitable. Therefore, according to another aspect of the present disclosure, the second kinematic information is information about the magnitude of acceleration in the target time unit, or the direction of the velocity in the previous time unit of the target time unit and the velocity in the target time unit. It may include at least one of information about the amount of angular change between directions.

20 shows an example of second kinematic information including information about acceleration.

As exemplarily described above in the present description, when kinematic information is expressed based on an entity-based coordinate system, even if an estimated load index is determined using only acceleration information, a motion load according to a difference between an orientation and a movement direction of an entity. Differences in degree may be reflected. Accordingly, the second kinematic information may include acceleration information (a1, a2, a3) of the target entity. Acceleration information may be expressed based on various coordinate systems, for example, acceleration information based on longitudinal, width, and height coordinates (a1), acceleration information based on orientation, lateral, and vertical coordinates (a2), At least one of acceleration information (a3) based on forward, backward, leftward, rightward, upward, and downward coordinates may be included in the second kinematic information as the acceleration information.

Meanwhile, when kinematic information is expressed based on a stadium-based coordinate system, for example, a difference between an orientation and a movement direction of an entity cannot be reflected in determining an exercise load using only acceleration information. In this case, the second kinematic information may include information on magnitude of acceleration (a4, a5, a6, a7) and motion direction change information (ad1, ad2, ad3, ad4). The acceleration magnitude information can be expressed based on various coordinate systems, for example, information on the magnitude of acceleration based on orientation direction, lateral direction, and vertical direction coordinates (a4), and acceleration based on forward, backward, lateral, and vertical direction coordinates. Magnitude information (a5), horizontal acceleration, which is magnitude information for the sum of the longitudinal acceleration and transverse acceleration, magnitude information (a6), and the sum of the magnitude information for the sum of longitudinal, transverse, and height-direction accelerations At least one of the acceleration magnitude information (a7) may be included in the second kinematic information as acceleration magnitude information. In addition, motion direction change information may also be expressed based on various coordinate systems, and may be determined from, for example, previously sampled orientations. For example, a horizontal acceleration direction (ad1), which is an angular difference between an acceleration direction of a previous time unit and an acceleration direction of a current time unit, may be utilized as movement direction change information. Alternatively, the magnitude (ad2) of the angular difference between the acceleration direction of the previous time unit and the acceleration direction of the current time unit may be utilized as motion direction change information. According to another aspect, the movement direction change may be determined based on the change in velocity direction, and more specifically, the horizontal movement direction (ad3), which is the angular difference between the velocity direction of the previous time unit and the velocity direction of the current time unit, is It can be used as movement direction change information. Alternatively, the magnitude (ad4) of the angular difference between the velocity direction of the previous time unit and the velocity direction of the current time unit may be utilized as motion direction change information.

Further, according to one aspect, the information about the linear motion may further include third kinematic information 1625 related to the jerk of the target entity. According to one aspect of the present disclosure, the third kinematic information is information on the magnitude of jerk in the target time unit, or between the direction of the jerk in the previous time unit of the target time unit and the direction of the jerk in the target time unit. At least one of information about the amount of angular change may be included.

Similar to acceleration, the magnitude of the jerk in information about the jerk utilizes the magnitude of the jerk in the target time unit, but in the information to reflect the relationship between the direction of orientation and the direction of motion of the target entity, it is acceleration rather than the change in direction of the jerk. A change in direction of may be more appropriate, and a change in direction of velocity may be more appropriate than a change in direction of acceleration. Therefore, according to another aspect of the present disclosure, the third kinematic information is information about the magnitude of jerk in the target time unit, or the direction of acceleration in the previous time unit of the target time unit and the acceleration in the target time unit. It may include at least one of information about the amount of angular change between directions. According to another aspect, the third kinematic information is information about the magnitude of jerk in the target time unit, or an angular change amount between a direction of velocity in a time unit previous to the target time unit and a direction of velocity in the target time unit. It may include at least one of information about.

21 shows an example of third kinematic information including jerk-related information.

As previously described as an example in the present description, when kinematic information is expressed based on an entity-based coordinate system, even if an estimated load index is determined using only jerk information, a motion load according to a difference between an orientation and a movement direction of an entity. Differences in degree may be reflected. Accordingly, the third kinematic information may include jerk information (j1, j2, j3) of the target entity. Jerk information can be expressed based on various coordinate systems, for example, jerk information (j1) based on longitudinal, width, and height coordinates, jerk information (j2) based on orientation, lateral, and vertical coordinates, At least one of forward, backward, leftward, rightward, upward, and downward coordinate-based jerk information (j3) may be included in the third kinematic information as jerk information.

Meanwhile, when kinematic information is expressed based on, for example, a stadium-based coordinate system, a difference between an orientation and a movement direction of an entity cannot be reflected in determining an exercise load using only jerk information. In this case, the third kinematic information may include jerk size information (j4, j5, j6, j7) and motion direction change information (jd1, jd2, jd3, jd4, jd5, jd6). The jerk size information can be expressed based on various coordinate systems. For example, the jerk size information (j4) based on orientation direction, lateral direction, and vertical direction coordinates, and jerk size information based on forward, backward, lateral, and vertical direction coordinates. Size information (j5), sum of lengthwise jerk and widthwise jerk Horizontal jerk, which is size information for jerk Size information (j6), sum of lengthwise jerk, widthwise jerk, and heightwise jerk Total sum of size information for jerk At least one of the jerk size information (j7) may be included in the third kinematic information as jerk size information. In addition, motion direction change information may also be expressed based on various coordinate systems, and may be determined from, for example, previously sampled orientations.

For example, a horizontal jerk direction (jd1), which is an angular difference between a jerk direction of a previous time unit and a jerk direction of a current time unit, may be utilized as motion direction change information. Alternatively, the magnitude (jd2) of the angular difference between the jerk direction of the previous time unit and the jerk direction of the current time unit may be utilized as jerk direction change information. According to another aspect, the movement direction change may be determined based on the change in the acceleration direction, and more specifically, the horizontal acceleration direction (jd3), which is the angular difference between the acceleration direction of the previous time unit and the acceleration direction of the current time unit, moves It can be used as direction change information. Alternatively, the magnitude (jd4) of the angular difference between the acceleration direction of the previous time unit and the acceleration direction of the current time unit may be utilized as motion direction change information. According to another aspect, the movement direction change may be determined based on the change in the velocity direction, and more specifically, the horizontal movement direction (jd5), which is the angular difference between the velocity direction of the previous time unit and the velocity direction of the current time unit, is It can be used as movement direction change information. Alternatively, the magnitude (jd6) of the angular difference between the velocity direction of the previous time unit and the velocity direction of the current time unit may be utilized as the movement direction change information.

Referring back to FIG. 16 , kinematic information 1600 may include information 1630 on angular movement. Even though the position of the target entity does not change in the exercise session, a case in which the target entity performs a rotational motion may occur. For example, if the target entity changes its orientation direction and performs an action to maintain possession of the ball in order to respond to the pressure defense of the opposing team player in possession of the ball in a soccer game, if only linear motion is included as kinematic information Such a change in the orientation direction does not have any effect on the exercise load. In addition, when performing a Cruyff turn in a dribbling state, it may be determined that the dribble performing the Cruyff turn and the dribbling not performing the same exercise load. According to one aspect of the present disclosure, the kinematic information includes information 1630 for each movement so that the exercise load generated by the rotational motion of the body of the target entity without a change in position is also reflected in the determination of the estimated load index. can do. For example, the kinematic information according to one aspect of the present disclosure includes information on at least one of a rotational movement in a roll direction, a rotational movement in a pitch direction, or a rotational movement in a yaw direction of a target entity. and may include information on at least one of angular velocity, angular acceleration, or each jerk of the target entity.

In this regard, as described above in this description, at least one of an inertial coordinate system and a hybrid coordinate system may be used as coordinates for each motion.

22 shows an example of each motion information based on an inertial coordinate system. As shown in FIG. 22, the inertial coordinate system-based angular motion information 1631 includes at least one of angular velocity information (av1), angular acceleration information (aa1), and angular jerk information (aj1) for the rotational motion of the target entity. can include The angular velocity information (av1) may include angular velocity based on a roll axis, angular velocity based on a pitch axis, and angular velocity based on a yaw axis, and angular acceleration information (aa1) may include angular acceleration based on a roll axis, angular velocity based on a pitch axis It may include the angular acceleration of the reference and the angular acceleration of the yaw axis. Each jerk information aj1 may include each jerk based on a roll axis, each jerk based on a pitch axis, and each jerk based on a yaw axis.

23 shows an example of each motion information based on a hybrid coordinate system. For example, in a team sports event such as soccer, rotational motions of players may occur mainly on a vertical axis, whereas rotational motions on the other axes may be thought to be insignificant. The hybrid coordinate system can consider only rotational motion based on a vertical axis as a rotational motion. As shown in FIG. 23, the hybrid coordinate system-based angular motion information 1633 includes at least one of angular velocity information (av2), angular acceleration information (aa2), and angular jerk information (aj2) for the rotational motion of the target entity. can include The angular velocity information (av2) may include angular velocity based on the vertical axis, the angular acceleration information (aa2) may include angular acceleration based on the vertical axis, and each jerk information (aj2) may include angular velocity based on the vertical axis. Each jerk can be included.

It should be noted that kinematic information according to embodiments of the present disclosure may be utilized in various combinations including at least some of the various elements described above.

State information-based exercise load determination

24 is a schematic flowchart of a method for providing exercise load information according to an embodiment of the present disclosure. FIG. 25 is a detailed flowchart of the estimated load index determining step of FIG. 24 . Hereinafter, a method for providing exercise load information according to an embodiment of the present disclosure will be described in more detail with reference to FIGS. 24 and 25 .

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 server 1500 as described with reference to FIGS. 1 to 3 or the controller 1520 included in the server 1500, and includes a processor and a memory. Any computing capable device may be used, and a combination of a plurality of physically separate devices may be referred to as a computing device.

A method for providing exercise load information according to an embodiment of the present disclosure determines an estimated load index by further considering state information reflecting characteristics related to a target exercise session as well as kinematic information during a target exercise session of a target entity As a result, information on the degree of exercise load can be determined with more improved accuracy. For example, even if the same intensity of training is performed, the exercise load felt by the athlete may be different due to various factors such as the athlete's condition on the day of training or environmental factors such as weather. According to one aspect of the present disclosure, by estimating an estimated load index by reflecting not only information about the motion of the target entity, but also various additional information that may affect the degree of exercise load of the target entity, Even the degree of exercise load can be reflected.

In this regard, the exercise load information providing method according to an embodiment of the present disclosure may determine an estimated load index based on “activity information”. Here, activity information may include kinematic information as described above in this description and condition information reflecting characteristics related to a target exercise session. Activity information may include information detected by the sensing platform 1100 as described above in this description, and may be understood as a concept encompassing any information that may be used to determine the degree of exercise load.

26 illustratively illustrates activity information according to an aspect of the present disclosure. As shown in FIG. 26 , activity information 2600 includes kinematic information 1600, session condition information 1610, and physical condition information 2620 as described above. may contain at least one. Here, the body condition information may include at least one of heart rate information 2621 , body temperature information 2623 , and respiration information 2625 . Information on heart rate can be obtained using indicators such as %HR _max (HR _act /HR _max ), %HRR ((HR _act - HR _rest ) / (HR _max - HR _rest )), and %VO _2max (VO _2act / VO _2max ). can include State information may be understood as a concept including at least one of session state information and body state information. Examples of more diverse state conditions are disclosed in detail below in this description.

As shown in FIG. 24 , the exercise load information providing method according to an embodiment of the present disclosure includes acquiring a target data set including activity information (S2410) and determining an estimated load index (S2420). can include

Hereinafter, each step of an embodiment of a method for providing exercise load information will be described.

As shown in FIG. 24 , the computing device may obtain a target data set including activity information (S2410). More specifically, the computing device may obtain a target data set for a target entity during a target workout session comprising a plurality of time units, wherein the target data set includes a target entity for each of the plurality of time units. In addition to the sequence of kinematic information of , it may include condition information that reflects characteristics related to the target exercise session. Here, the state information may be obtained for each of a plurality of time units included in the target exercise session similarly to kinematic information and may be configured to have a sequence of state information, or one state information for one exercise session. It may be configured to have. Alternatively, some of the plurality of state information may be obtained for each time unit to form a sequence, and other state information may have one state information for an exercise session. State information is detailed later in this description.

Referring back to FIG. 24 , the computing device may determine an estimated load index (S2420). More specifically, the computing device may determine, from the target data set, using an artificial neural network, an estimated load index that reflects the degree of exercise load of the target exercise session for the target entity. According to one aspect, the determined estimated load index may be provided as an estimate of exercise load for a target exercise session. Here, the estimated load index may include, for example, a plurality of numerical values indicating the degree of exercise load, and may be configured such that 10 represents the highest exercise intensity with a value of 1 to 10, for example, but in this specific numerical range Not limited.

An estimated load index according to an aspect of the present disclosure may be an estimate of a rate of perceived fatigue (RPE) of a target entity for a target exercise session. For example, an estimated load index determined according to an aspect of the present disclosure may be provided instead of the perceived fatigue as exemplarily illustrated in FIG. 4 .

Relatedly, FIG. 25 is a detailed flowchart of the estimated load index determining step of FIG. 24 . As shown in FIG. 25, the computing device first prepares a trained artificial neural network (S2421), inputs a target data set to the input layer of the artificial neural network (S2423), and based on the result value of the output layer of the artificial neural network It is possible to obtain an estimated load index (S2425).

Here, the artificial neural network may be trained using a training data set labeled with an exercise load index to output an exercise load index of the target exercise session when activity information of a target entity for the target exercise session is input. Basically, the learning data set may be generated from activity information about a target entity that has performed an exercise session, and an exercise load index to be labeled may be a value collected from the target entity that has performed an exercise session.

More specifically, but not limitedly, the artificial neural network according to an embodiment of the present disclosure may be trained based on a plurality of training data sets including the first training data set. Here, the first training data set includes a sequence of kinematic information of the first entity during the first workout session and state information about the first workout session, and the first training data set includes a sequence of kinematic information for the first workout session and an exercise load of the first workout session from the first entity It may be labeled as a collected score.

The estimated load index determination process using an artificial neural network may proceed similarly to the process described above with reference to FIGS. It may be slightly different in that it includes.

Hereinafter, state information according to embodiments of the present disclosure will be described in more detail. As shown in FIG. 26 , the state information may include session state information 2610 for characteristics of the target exercise session itself for which the estimated load index is to be determined and information 2620 for the state of the target entity itself. For the same training program, a player may feel a greater exercise load if the condition of the day is not good, and even if the same training program is performed in harsh climate conditions such as high temperature and humidity, a greater exercise load may be felt.

According to one aspect of the present disclosure, status information may include daily readiness response information collected from a target entity for a target exercise session. In this regard, according to the trend of operating modern sports clubs, all players participating in the exercise session preceding or following each exercise session often go through the process of investigating and recording the daily readiness at the time of participation in the exercise session. there is. A survey on daily readiness like this is also referred to as a wellness survey. The wellness survey may include various queries related to the health or condition of the target entity participating in the exercise session.

The daily readiness response information according to one aspect includes a sleep time evaluation value, a sleep quality evaluation value, a fatigue level evaluation value, a soreness level evaluation value, and a stress level evaluation value. level) evaluation value or mood evaluation value, but is not limited thereto, and arbitrary evaluation value information related to the readiness of the target entity may be included.

27 is an example of a wellness survey for securing daily readiness response information according to an embodiment of the present disclosure. As exemplarily illustrated in FIG. 27 , a plurality of items for daily readiness evaluation may be queried, and for example, an evaluation value may be received for each inquiry. Evaluation values of each collected item may be included in status information. The state information may include evaluation values for each item in parallel or may include the sum of evaluation values for each item.

Even for the same amount of exercise, the degree of exercise load may be different depending on psychological factors (psychological stress, mental factor) of a target entity such as a sports player. Psychological factors that may affect the degree of exercise load may include factors of the target entity itself, or may include various conditions or state factors of the target exercise session itself.

In this regard, according to one aspect of the present disclosure, state information may include exercise type information. For example, the exercise type information may reflect information on whether a target exercise session is a match session or a training session. The workout session may include a game session in which results such as a win or loss are recorded and league rankings or tournament advances are determined, and a practice session in which training is performed in preparation for the game session. Compared to the practice session, in the game session, even if the same amount of exercise is obtained, a greater exercise load may be felt due to psychological factors. According to one aspect of the present description, by including exercise type information such as information on whether the status information is a game session or a practice session, for example, considering the difference in the degree of exercise load according to the type of the target exercise session It is possible to determine the estimated load index.

Also, according to one aspect of the present disclosure, the status information may include an importance evaluation value according to a type of a target exercise session. Even if they experience the same amount of exercise, if the importance of a corresponding exercise session is higher, a target entity such as a sports player may have a greater psychological burden and consequently experience a greater degree of exercise load. For example, as described above, the importance of a game session may be greater than that of a practice session, and a target entity may experience a greater degree of exercise load even with the same amount of exercise due to a high psychological load on the target entity. In addition, in the case of rival matches where attention is focused, such as 'Korea-Japan match' or 'El Clasico', psychological burden may be much higher than that of general match sessions. Alternatively, when an important game in a tournament such as the 'final match' or a special meaning such as a 'league championship match' has a high importance and can cause a great psychological burden. The importance according to the type of target exercise session may be reflected as an evaluation value by reflecting such various factors. The importance evaluation value may be expressed as a numerical value such as 1 to 5, for example, or may be classified into an importance type and included in the training data set and the target data set, respectively.

According to one aspect of the present disclosure, the status information may include information about the length of time elapsed from the previous game session to the target workout session. According to modern sports science, a training schedule is designed to exhibit optimal performance in a game session, and the state of a target entity participating in a workout session may be different according to an elapsed time from the game session. For example, the exercise load felt by the sports player on the second day training and the third day training after the game session may be different even for the same exercise amount. Therefore, state information according to one aspect of the present disclosure includes information on the length of time that has elapsed from the previous game session to the target exercise session, and is estimated by reflecting the difference in the degree of exercise load according to the elapsed time from the game session. load index can be determined.

According to one aspect of the present disclosure, the status information may include information about a length of time that has elapsed from a start date of a sports season including a target exercise session to the target exercise session. In general, team sports events are conducted over a long period of time. Various conditions such as physical fitness or physical condition of a target entity such as a sports player participating in an exercise session may fluctuate while digesting a long season, ranging from a minimum of several months to a year. For example, as the season progresses, the target entity may experience a relatively high degree of exercise load even for the same amount of exercise. Conversely, at the very beginning of the season, despite training for the season, you may feel a high degree of exercise load for the same amount of exercise for the same reason as adapting to actual combat. According to one aspect of the present disclosure, a target exercise session for which the degree of exercise load is to be determined includes information on the length of time that has elapsed from the start date of the sports season in which the target exercise session is included, so that the target exercise session is set at the beginning of the season. The estimated load index may be determined by reflecting the progress of the season, such as whether it is mid-season, mid-season, or late season.

According to one aspect of the present description, condition information may include information about a physical condition of a target entity. Even for the same amount of exercise, a target entity such as a sports player may feel a different degree of exercise load depending on what physical state he or she is in. Accordingly, according to one aspect of the present disclosure, the estimated load index may be determined by reflecting the condition information including information on the physical condition of the target entity.

Here, the information on the physical condition may include at least one of age, height, weight, physical fitness evaluation value, muscular strength evaluation value, maximum heart rate, or maximum oxygen intake. However, the body state is not limited thereto, and various evaluation factors that may indicate the body state of the target entity may be included in the body state information.

According to one aspect of the present disclosure, state information may include environment information about a target exercise session. For example, the environment information may reflect at least one of weather information, temperature information, humidity information, or season information, but is not limited thereto. As described above, when the weather is very hot or the humidity is high, fatigue may be felt even for the same amount of exercise. In addition, an exercise session performed in rainy weather may cause a relatively large degree of exercise load. Accordingly, according to one aspect of the present disclosure, the estimated load index may be determined by reflecting environmental factors of the target exercise session by allowing the state information to include environment information of the target exercise session.

According to one aspect of the present disclosure, state information may include biological measurement information of a target entity during a target exercise session. Here, the biometric information may include at least one of heart rate information, information on the ratio of heart rate to maximum heart rate, body temperature information, information on oxygen intake, or information on the ratio of oxygen intake to maximum oxygen intake. can As described above with reference to FIG. 26 , the body state information may include biological measurement information among body state information, such as heart rate information, body temperature, and respiration state information. Even for the same amount of exercise, the felt exercise load may be different according to the biological state of the target entity. According to one aspect of the present description, by including such biological information as additional information in state information, it is possible to determine an estimated load index by more accurately reflecting different exercise load levels according to target entities. Here, the biological measurement information may be obtained through various sensors such as the biometric sensing module 1317 provided in the sensing platform.

Although various activity information has been described above, the determination of the estimated load index according to the exercise load information providing method according to an embodiment of the present disclosure is a combination of at least some of the above-described plurality of activity information, kinematic information, and state information. It should be noted that it may be performed based on a target data set according to

Calibration of estimated load index

28 is a schematic flowchart of a method for providing exercise load information according to an embodiment of the present disclosure. 29 is a detailed flowchart of an embodiment of the step of determining the estimated load index of FIG. 28; FIG. 30 is a detailed flowchart of another embodiment of the step of determining the estimated load index in FIG. 28; Hereinafter, a method for providing exercise load information according to an embodiment of the present disclosure will be described in more detail with reference to FIGS. 28 to 30 .

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 server 1500 as described with reference to FIGS. 1 to 3 or the controller 1520 included in the server 1500, and includes a processor and a memory. Any computing capable device may be used, and a combination of a plurality of physically separate devices may be referred to as a computing device.

A method for providing exercise load information according to an embodiment of the present disclosure further improves the accuracy of the estimated load index by further reflecting the characteristics of the target entity in providing information on the degree of exercise load according to the target exercise session of the target entity. can make it According to an embodiment of the present disclosure, a target entity may adaptively determine an estimated load index by using "target-specific information" in determining the estimated load index.

In the present disclosure, “target-specific information” may be understood to include characteristics possessed by a target entity for which exercise load information is to be provided, or information for identifying such a target entity. For example, since the degree of exercise load experienced for the same amount of activity may be different depending on the exercise capacity of the target entity, the target-specific information may include the exercise capacity of the target entity as a characteristic of the target entity. Alternatively, information capable of evaluating the level of fitness, such as the affiliated league of the target entity, may be included in target-specific information as identification information of the target entity.

According to one aspect of a method for determining an adaptive exercise load of a target entity using target-specific information, a normal estimated load index determined based on kinematic information of a target entity is calibrated by applying an adjustment factor considering target-specific information. An estimated load index can be determined.

According to another aspect of the present disclosure, in learning an artificial neural network, for example, by using a network fusion structure, a target entity adaptive estimated load index is obtained by utilizing data considering characteristics of a target entity and typical data together. can be obtained

According to another aspect of the present disclosure, training data is classified according to a classification to which a target entity belongs, an artificial neural network is separately prepared for each classification, the artificial neural network is trained based on the training data corresponding to each, and an estimated load index is calculated. In determining, the adaptive estimated load index of the target entity may also be acquired by using the corresponding learned artificial neural network.

Hereinafter, embodiments for providing target entity adaptive exercise load information according to the present description will be described in more detail.

As shown in FIG. 28, the exercise load information providing method according to an embodiment of the present disclosure includes obtaining target-specific information (S210), obtaining a target data set (S2820), and an estimated load index. A step of determining (S2830) may be included.

Hereinafter, each step of an embodiment of a method for providing exercise load information will be described.

As shown in FIG. 28 , the computing device may acquire target-specific information (S2810). More specifically, the computing device can obtain target-specific information reflecting at least one of identification information or characteristic information about the target entity.

As discussed above, the target-specific information that can be used to determine the estimated load index adaptively to the target entity may include information unique to the target entity, and may include characteristic information representing the characteristics of the target entity itself, or , may include identification information for identifying the classification or type of the target entity. The target-specific information may include an adjustment factor that is retroactively applied to an estimated load index that can be determined without considering the target-specific characteristics to reflect the target-specific characteristics, and is input to the artificial neural network along with kinematic information. and may include data for determining the estimated load index. The target-specific information may be obtained by inputting from previously transmitted information, for example, or may be calculated through a specific process associated with the target entity.

Referring back to FIG. 28 , the computing device may obtain a target data set (S2820). More specifically, the computing device may obtain a target data set for a target entity during a target workout session that includes a plurality of time units, wherein the target data set includes a target data set of the target entity for each of the plurality of time units. It may contain sequences of kinematic information. At least some of the various processes previously described in this description may be applied to the acquisition procedure for the sequence of kinematic information.

Referring to FIG. 28 , the computing device may determine an estimated load index using an artificial neural network (S2830). More specifically, the computing device can determine, from the target-specific information and the target data set, using an artificial neural network, an estimated load index that reflects the degree of exercise load of the target exercise session for the target entity. . According to one aspect, the determined estimated load index may be provided as an estimate of exercise load for a target exercise session. Here, the estimated load index may include, for example, a plurality of numerical values indicating the degree of exercise load, and may be configured such that 10 represents the highest exercise intensity with a value of 1 to 10, for example, but in this specific numerical range Not limited.

An estimated load index according to an aspect of the present disclosure may be an estimate of a rate of perceived fatigue (RPE) of a target entity for a target exercise session. For example, an estimated load index determined according to an aspect of the present disclosure may be provided instead of the perceived fatigue as exemplarily illustrated in FIG. 4 .

Relatedly, FIG. 29 is a detailed flowchart of an embodiment of the estimated load index determining step of FIG. 28 . According to an embodiment, the determination of the estimated load index consists of determining the estimated load index based on kinematic information without considering the characteristics of the target entity and applying an adjustment factor considering the characteristics of the target entity to the determined estimated load index. to determine the target entity's adaptively enhanced accuracy estimated load index.

More specifically, as shown in FIG. 28, the computing device first prepares a trained artificial neural network (S2831) and inputs a target data set to the input layer of the artificial neural network (S2833), so that the output layer of the artificial neural network An estimated load index may be obtained based on the result value (S2835).

Here, the artificial neural network may be trained using a training data set labeled with an exercise load index to output an exercise load index of the target exercise session when activity information of a target entity for the target exercise session is input. Basically, the learning data set may be generated from activity information about a target entity that has performed an exercise session, and an exercise load index to be labeled may be a value collected from the target entity that has performed an exercise session.

More specifically, but not limitedly, the artificial neural network according to an embodiment of the present disclosure may be trained based on a plurality of training data sets including the first training data set. Here, the first training data set may include a sequence of kinematic information of the first entity during the first exercise session, and may be labeled with a score collected from the first entity for an exercise load of the first exercise session. .

At least some of the processes described above with reference to FIGS. 6 to 8 in relation to providing exercise load information based on kinematic information may be applied to the estimated load index determination process using an artificial neural network.

For example, as shown in FIG. 29 , the computing device inputs a target data set to an input layer of an artificial neural network (S2833) and obtains an estimated load index based on a result value of an output layer of the artificial neural network (S2835). there is.

However, here, according to one aspect of the present disclosure, target-specific information may include a modifier for an estimated load index corresponding to a target entity. That is, information for determining a calibrated estimated load index with improved accuracy by reflecting the unique characteristics of the target entity may be included. In this case, obtaining the estimated load index (S2835) may be to obtain the estimated load index by multiplying the resultant value of the output layer of the artificial neural network by an adjustment factor. Accordingly, the obtained estimated load index may become an estimated value of an exercise load closer to what the target entity actually feels by reflecting the unique characteristics of the target entity.

Hereinafter, examples of adjustment factors according to embodiments of the present disclosure will be described in more detail.

According to one aspect, the adjustment factor may reflect fitness evaluation information of the target entity. More specifically, the adjustment factor may reflect the ratio of the game load index to the maximum value of the load index, wherein the game load index is based on kinematic information during at least one recent game session of the target entity. It may be obtained using an artificial neural network.

In this regard, the adjustment factor may be for evaluating the degree of exercise load by reflecting the individual characteristics (individuality) of the target entity. For example, if an artificial neural network for determining an estimated load index is trained based on a learning data set obtained from a plurality of entities, it is assumed that the fitness levels of the plurality of entities are the same. In general, considering the degree of difference in athletic ability between sports players within the same level of league in which learning data is collected, according to one embodiment, even if it is assumed that the athletic ability of a plurality of entities is the same, it is necessary to calculate the estimated load index. There may not be any serious errors. However, there is a difference in exercise capacity among individual sports players, and in order to more accurately estimate the degree of exercise load, it is more advantageous to determine the estimated load index by reflecting the exercise ability of each sports player. For the same amount of exercise, an entity with high exercise capacity will experience a relatively low exercise load, and an entity with low exercise capacity will experience a relatively high exercise load, so the estimated load index can be appropriately calibrated based on the exercise capacity.

As a factor for reflecting the athletic ability of the target entity, an estimated load index of an estimated load index for a game session, that is, a game load index may be considered. The exercise session may be divided into a game session, which is a session in which an actual game is performed, and a practice session in which practice is performed for this.

In a team sports game, all players perform an exercise by exerting their maximum physical strength or athletic ability, and when investigating the perceived fatigue of the response method, all players have a maximum value of In the BORG method, a 10-point RPE) is answered. Considering this point, the ratio between the estimated load index calculated by inputting the kinematic information of the target entity collected for the game session into the artificial neural network and the maximum value (eg, 10) of the load index is reflected to calculate the target entity It is possible to determine the adjustment factor reflecting the exercise ability of

31 is an exemplary diagram for determining a game load index. As shown in FIG. 31 , a game load index 3130 may be determined based on a value output by inputting kinematic information 3110 during at least one recent game session of a target entity to an artificial neural network 3120. Here, the artificial neural network 3120 may utilize an artificial neural network for determining an estimated load index before reflecting the adjustment factor of the target entity, or may be provided with and utilized a separate learned artificial neural network. The game load index 3130 may be used as the game load index when an estimated load index is determined using, for example, kinematic information for a recent game. Alternatively, a plurality of estimated load indices may be determined using kinematic information on a plurality of recent games, and a representative value, for example, an average value of the plurality of estimated load indices may be used as the game load index.

For a predetermined first target entity, a game load index determined based on kinematic information about a game session may be a value smaller than the maximum value of the load index. The response load index obtained by querying the first target entity for the match session will obtain a full score (eg 10 points), but the match load index for the first entity based on the artificial neural network will be 8 points, for example. can be determined Accordingly, for the first entity, an adjustment factor reflecting the ratio (8:10) of the maximum value of the load index and the game load index may be determined. For example, in the simplest example, the estimated load index of the first target entity may be determined by multiplying a result value output by an artificial neural network based on a target data set of a target exercise session by 10/8 as an adjustment factor. When the output value of the artificial neural network for the first target entity determined based on the target data set of the target exercise session is 7, the estimated load index of the first target entity reflecting the adjustment factor may be determined to be 8.75 (7 * 10 / 8). there is.

However, this is only an exemplary method of determining the adjustment factor, and various calibration criteria may be calculated based on the ratio of the maximum value of the load index and the game load index.

According to one aspect of the present invention, the adjustment factor may reflect evaluation information of the affiliated league of the target entity. More specifically, the adjustment factor may reflect the ratio of the evaluation value of the affiliated league of the target entity to the evaluation value of the affiliated league of the first entity in the training data set.

As described above, a difference in athletic ability between a plurality of sports players belonging to the same league may not be large. However, expected athletic abilities may differ significantly depending on the league to which the target entity belongs, such as, for example, a professional league, an amateur league, and a youth league. In this regard, an adjustment factor may be determined by reflecting evaluation information of a league corresponding to learning data used to train an artificial neural network and evaluation information of a league to which a target entity belongs.

For example, if the artificial neural network used for determining the estimated load index is learned based on kinematic information of target entities in a professional league, while the target entity is a sports player belonging to a youth league, professional league evaluation information (eg For example, 10 points) and youth league evaluation information (for example, 6 points), an adjustment factor of 10/6 can be determined as a simple example. That is, if the target data set for the target exercise session of the second target entity belonging to the youth league is input to the artificial neural network learned based on the training data set corresponding to the professional league and the output value is 5, the adjustment factor is reflected. 2 The estimated load index of the target entity can be determined as 8.34 (5 * 10 / 6). However, this is only an exemplary method for determining an adjustment factor, and various calibration criteria may be calculated based on a ratio of evaluation information of a rig corresponding to a training data set and evaluation information of a rig corresponding to a target entity.

Meanwhile, FIG. 30 is a detailed flowchart of another embodiment of the step of determining the estimated load index of FIG. 28 . According to an embodiment, the estimated load index determination is performed by determining the estimated load index based on an artificial neural network in consideration of both kinematic information and characteristic information of the target entity, thereby adaptively improving the estimated load index of the target entity with improved accuracy. can be determined.

In this regard, according to one aspect of the present disclosure, in learning an artificial neural network, for example, by using a network fusion structure, data obtained in a game session and data obtained in a typical exercise session for considering the characteristics of a target entity data can be used together. An adaptive estimated load index of the target entity may be obtained by inputting the target data set during the target exercise session of the target entity and the data of the match session of the target entity together into the artificial neural network trained as described above.

More specifically, as shown in FIG. 30 , the computing device first prepares a trained artificial neural network (S2837), inputs a target data set and target-specific information to the input layer of the artificial neural network (S2838), and An estimated load index may be obtained based on the result value of the output layer of the neural network (S2839).

Here, the artificial neural network outputs an exercise load index of the target exercise session when the activity information of the target entity for the target exercise session and the activity information for the game session of the target entity are input together. A training data set labeled with the exercise load index, and It may be learned using a match data set labeled with the maximum score of the exercise load index.

A learning data set for training the artificial neural network may be generated from, for example, activity information about an entity that has performed an exercise session, and an exercise load index to be labeled may be a value collected from an entity that has performed an exercise session. In addition, a game data set for training an artificial neural network may be generated from activity information of at least one game session of an entity performing the exercise session, and may be labeled as a maximum value of an exercise load index.

More specifically, but not limitedly, the artificial neural network according to an embodiment of the present disclosure may be trained based on a plurality of training data sets each including a first training data set and a first game data set.

Here, the first training data set may include a sequence of kinematic information of the first entity during the first exercise session, and may be labeled with a score collected from the first entity for an exercise load of the first exercise session. .

In addition, the first game data set may include a sequence of kinematic information of the first entity during at least one game session, and may be labeled with a maximum score.

According to one aspect, the artificial neural network may be trained using a fusion of a training feature map corresponding to the first training data set and a game feature map corresponding to the first game data set. For example, a structure for training an artificial neural network by using two or more different types of feature values as learning data, such as training an artificial neural network using RGB value-related data and depth-related data of an image, respectively. possible. For example, like a network fusion structure, an artificial neural network may include a first network for generating a first feature map for a first feature value and a second network for generating a second feature map for a second feature value. . A value output to the output layer may be determined by using a fusion of the first feature map and the second feature map.

According to one aspect of the present disclosure, for an entity such as, for example, a sports player, the first characteristic value may be kinematic information obtained from a game session, which is a session in which the entity participates in a sports game. For the kinematic information of the game session, the maximum value for the exercise load may be labeled. The second characteristic value may be kinematic information obtained from any exercise session in which the entity participates. Any workout session may include both a game session and a practice session, or may include only a practice session excluding the game session. A response score for an exercise load of an entity participating in a corresponding exercise session may be labeled as kinematic information for an exercise session.

A training feature map may be created based on a first training data set comprising a sequence of kinematic information for any workout session of the first entity, comprising a sequence of kinematic information during a competition session of the first entity. A game feature map may be created based on the first game data set. The training feature map and the game feature map may be concatenated and provided to a fully connected (FC) layer directly or through an additional layer. Various network fusion structures are possible depending on when the training feature map and game feature map are concatenated. In relation to this, FIG. 33 illustrates an Early fusion model in network fusion, FIG. 34 illustrates a Middle fusion model in network fusion, and FIG. 35 illustrates a late fusion model in network fusion. fusion) model. As shown in FIGS. 33 to 35, the time at which the data due to the game data set and the data due to the training data set are concatenated (that is, concatenation in the early layer or concatenation in the later layer) is early, middle, It may be determined according to at least one of various types of network fusion, such as late combining, and is not limited to a specific network model.

As shown in FIG. 30 , when the artificial neural network trained as described above is prepared (S2837), the computing device inputs a target data set and target-specific information to the input layer of the artificial neural network (S2838), and outputs the artificial neural network An estimated load index may be obtained based on the result value of the layer (S2839). Here, the target-specific information acquired in the acquiring step (S2810) and input to the artificial neural network (S2838) may include a sequence of kinematic information of the target entity during at least one recent game session.

More specifically, FIG. 32 is an exemplary view of determining an estimated load index using a game data set and a training data set. 30 and 32 , the computing device may input a target data set 3220 and target-specific information 3210 to an input layer of an artificial neural network 3230 (S2838). Target data set 3220 can include a sequence of kinematic information from a target athletic session of a target entity, as described above, and target-specific information 3210 includes kinematic information during at least one recent competitive session of the target entity. Information 3210 may be included.

As shown in FIGS. 30 and 32 , the computing device may acquire an estimated load index 3240 of the target exercise session of the target entity based on the result value of the output layer of the artificial neural network 3230 (S2839). . Therefore, a method for providing exercise load information according to an aspect of the present disclosure may determine an estimated load index according to kinematic information of a target exercise session by reflecting kinematic information in a game session of a target entity, and achieve more improved accuracy. The target entity may provide adaptive exercise load information.

According to another aspect of the present disclosure, learning data is classified or collected according to a classification to which a target entity belongs, an artificial neural network is separately prepared for each classification, the artificial neural network is trained based on the learning data corresponding to each, and the estimated load is estimated. Even in determining the index, the adaptive estimated load index of the target entity may be acquired by using the corresponding learned artificial neural network.

According to one aspect, the target-specific information acquired in the step of acquiring the target-specific information (S2810) includes identification information about a league to which the target entity belongs, and the step of determining the estimated load index (S2830) includes: Among a plurality of artificial neural networks, an artificial neural network trained based on a plurality of training data sets corresponding to a league to which the target entity belongs may be used.

More specifically, entity-specific characteristics, such as, for example, fitness level, may be different depending on the group to which the corresponding entity belongs. For example, a sports player belonging to a youth league and a sports player belonging to a professional league may differ from each other in various unique characteristics, including, for example, athletic ability, and may experience different degrees of exercise load even for the same amount of exercise. there is. Therefore, according to one aspect of the present disclosure, entities subject to exercise load measurement are classified into a plurality of groups according to a predetermined criterion, and an artificial neural network corresponding to each classification is formed using training data sets corresponding to each classification. The estimated load index may be determined by training and inputting the target data set for the target exercise session of the target entity to the artificial neural network corresponding to the target entity. Accordingly, the estimated value of the degree of exercise load can be determined with improved accuracy by reflecting the unique characteristics of the target entity.

For example, information for classifying entities into a plurality of groups may be affiliated leagues. According to one aspect, an artificial neural network for a youth league trained on the basis of training data sets for sports players included in a youth league and a pro trained based on training data sets for sports players included in a professional league An artificial neural network for each rig may be prepared. According to one aspect, an affiliated league identifier for a target entity may be obtained as target-specific information. Using this, an estimated load index for the target entity may be determined based on a value output by inputting a target data set for the target entity to an input layer of an artificial neural network corresponding to a league to which the target entity belongs.

Although various embodiments of methods for providing exercise load information have been described as examples in the present description, at least some of the features of each embodiment may be applied to other embodiments as well. For example, even when target-specific information is used to consider the characteristics of a target entity, the target data set may include condition information in addition to kinematic information. That is, it should be understood that a method for determining exercise load information in which partial features of the embodiments of the present disclosure are mutually combined are also included within the scope of the technical idea of the present disclosure.

Entity Monitoring and Early Risk Detection

In the embodiments of the present description, the load index that can reflect the evaluation value for the degree of exercise load may include an “estimated load index” and a “response load index”. Here, the “estimated load index” may be an estimated value for an exercise load calculated based on activity information of a target entity during a target exercise session. For example, the estimated load index may be determined using an artificial neural network. The “response load index” may be a direct response from the target entity of a degree of fatigue that the target entity evaluates itself as experienced by the target exercise session, such as, for example, a perceived fatigue rate (RPE).

According to one aspect of the present disclosure, monitoring and early risk detection for an entity such as a sports player is possible by comparing the estimated load index and the response load index. When the difference between the degree of exercise load predicted by the sports player and the degree of exercise load actually felt by the sports player satisfies a predetermined criterion, it may be determined that the corresponding sports player has an abnormal symptom.

Recently, in the operation of modern professional sports clubs, a system for submitting the RPE after the end of each exercise session has been established. If there is a discrepancy between the predicted RPE for a particular player and the player's submitted RPE, this may indicate that the player's conditioning is experiencing problems. Therefore, it is possible to perform management for individual players at the club level by detecting that a significant difference exists between the RPE prediction and submitted data for a predetermined period, for example, weekly. In addition, even if there is a significant difference between the prediction of the representative RPE for a particular team and the submitted data, a review of whether problems occur due to factors outside the training program, such as inadequate meals or facilities, and heavy non-training schedules. can be made to do.

36 is a conceptual diagram of a method for providing exercise load information for entity monitoring and early risk detection according to an embodiment of the present disclosure. As shown in FIG. 36 , an estimated load index 3630 may be calculated by collecting activity information during an exercise session for a target entity 3610 to be monitored and inputting the information to an artificial neural network 3620 . In addition, by performing the survey procedure 3640 on the degree of exercise load on the target entity 3610, a response load index 3645 reflecting the self-assessment value for the corresponding exercise session can be obtained. By comparing (3650) the estimated load index 3630 and the response load index 3650, a notification message 3660 for reporting an abnormality of the target entity is generated when the difference between the two satisfies a predetermined condition. It can be created and passed on to the administrator. Here, the criterion for the difference between the estimated load index and the response load index is a determination that the difference between the two occurs continuously (eg, more than a week) or significantly (the difference between the estimated load index and the response load index is greater than a threshold value). can include In this case, it may be determined that the target entity is not in a normal state, and a corresponding action may be performed at the manager level.

37 is a schematic flowchart of a method for providing exercise load information according to an embodiment of the present disclosure, FIG. 38 is a detailed flowchart of an estimated load index calculation step of FIG. 37, and FIG. 39 is an estimated load index determination of FIG. 38 Here is a detailed flow chart of the steps. Hereinafter, a method of providing exercise load information according to an embodiment of the present disclosure will be described in more detail with reference to FIGS. 37 to 39 .

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 server 1500 as described with reference to FIGS. 1 to 3 or the controller 1520 included in the server 1500, and includes a processor and a memory. Any computing capable device may be used, and a combination of a plurality of physically separate devices may be referred to as a computing device.

As shown in FIG. 37 , the exercise load information providing method according to an embodiment of the present disclosure includes calculating an estimated load index (S3710) and receiving a response load index. It may include step S3720 and outputting a notification message (S3730).

More specifically, referring to FIG. 37 , the computing device may calculate an estimated load index (S3710). More specifically, the computing device may include an estimated load index reflecting an estimated value of an exercise load degree of a target exercise session for a target entity based on kinematic information of the target entity during the target exercise session ( Estimated load index) can be calculated.

Here, the estimated load index may be understood as an estimated value of a rate of perceived fatigue (RPE) for a target exercise session of a target entity.

The process of calculating the estimated load index may be performed by reflecting at least some of the processes included in various embodiments as described above in this description. For example, calculating the estimated load index (S3710) may include obtaining a target data set (S3711) and determining the estimated load index (S3713), as shown in FIG. 38 . . More specifically, the computing device may obtain a target data set for a target entity during a target exercise session including a plurality of time units (S3711). Here, the target data set may include a sequence of kinematic information of a target entity for each of a plurality of time units. In addition, the computing device may determine an estimated load index for the target entity from the target data set using an artificial neural network (S3713). More specifically, as shown in FIG. 38 , the computing device may prepare a trained artificial neural network based on a plurality of training data sets including the first training data set (S3713a). Here, the first training data set may include a sequence of kinematic information of the first entity during the first exercise session and may be labeled with a score collected from the first entity for an exercise load of the first exercise session. In addition, the computing device may input the target data set to the input layer of the artificial neural network (S3713b) and obtain an estimated load index based on a result value of the output layer of the artificial neural network (S3713c).

Although an exemplary process for obtaining the estimated load index has been described above, the processes for obtaining various estimated load indices previously described in this description (e.g., activity information-based determination, calibration procedure, consideration of target entity-specific characteristics) etc.) may be combined to obtain an estimated load index.

Referring back to FIG. 37 , the computing device may receive a response load index reflecting a self-evaluation value for the degree of exercise load of a target exercise session by a target entity (S3720). can Here, the response load index may reflect the perceived fatigue of the target entity for the target exercise session. For example, after the target entity participates in the target exercise session, it may be made to go through a survey procedure on perceived fatigue. The target entity may respond with a numerical value between 1 and 10, for example, the degree of exercise load according to the target exercise session felt by the target entity, and the corresponding information may be input through a predetermined input device and provided to the computing device. That is, the response load index may reflect an evaluation value for perceived fatigue experienced by the actual target entity.

As shown in FIG. 37 , the computing device may output a notification message based on a comparison result between the estimated load index and the response load index (S3730). That is, if the difference between the estimated load index and the response load index satisfies a predetermined criterion, the target entity can be considered to be in an abnormal state, and the computing device informs the administrator whether or not the abnormal state has occurred by outputting a notification message. can

Calculating the estimated load index and receiving the response load index may be continuously performed for one or more exercise sessions. That is, according to one aspect, the target workout session may be a collection of a plurality of workout sessions. Accordingly, comparison between the estimated load index and the response load index may be performed for a plurality of exercise sessions. The criterion for the difference between the two may be set for a difference for one exercise session or whether or not the difference for a plurality of exercise sessions is sustained.

More specifically, according to one aspect of the present description, outputting a notification message (S3730) includes outputting a notification message in response to a determination that the difference between the estimated load index and the response load index is greater than or equal to a first predetermined threshold value. can be configured to That is, the computing device calculates a difference value between the estimated load index and the response load index respectively corresponding to a specific target exercise session, and if the calculated difference value is equal to or greater than a predetermined first threshold value (eg, 3), the target entity It can be judged to be in an abnormal state. For example, if the degree of exercise load actually felt by the target entity is too large compared to the degree of exercise load predicted based on motion data such as kinematic information, the target entity is highly likely to have a problem. Accordingly, the computing device may notify the manager by outputting a notification message.

According to another aspect of the present disclosure, the outputting of the notification message (S3730) may include determining whether a difference between the estimated load index and the response load index is greater than or equal to a predetermined second threshold for exercise sessions that have been continued for more than a predetermined first threshold number of times. It may be configured to output a notification message in response to the determination. For example, even if the difference between the estimated load index and the response load index for the target entity is not too large, if a difference of a certain level or more is repeated, there is some problem in the target entity or a major problem such as injury occurs. It can be judged as a precursor symptom of something. Accordingly, the computing device monitors the difference between the estimated load index and the response load index for a plurality of exercise sessions, and sets the difference between the estimated load index and the response load index to a predetermined second threshold value (eg, 1). When the number of exercise sessions having the above number of times (eg, 5 times) or more consecutively determined beforehand occurs, a manager may be notified of this by outputting a notification message.

According to another aspect of the present disclosure, the outputting of the notification message (S3730) may include a second exercise session in which a difference between an estimated load index and a response load index among a plurality of consecutive exercise sessions is equal to or greater than a predetermined third threshold value. and outputting a notification message in response to determining that the ratio is greater than or equal to a first predetermined threshold ratio. For example, when monitoring is performed on the difference between the estimated load index and the response load index for a plurality of exercise sessions, if a difference of a certain level or more occurs more often than an acceptable level even if it does not occur continuously, this is also the target It may also be interpreted that the entity is experiencing a particular problem or exhibits precursory symptoms of injury. Accordingly, the computing device may determine, for example, an exercise session showing a difference between estimated-load indices greater than or equal to a third threshold value (eg, 1.5) for a plurality of exercise sessions, for example, five or more consecutive exercise sessions. For example, when it occurs three or more times, it is determined that the ratio corresponds to a predetermined first threshold ratio (eg, 0.6) or more, and a notification message is output to notify the manager.

Exemplary abnormal state determination criteria have been described above, but the technical spirit of the present disclosure is not limited thereto, and various criteria for determining the degree of difference using the estimated load index and the response load index for the target entity are all of the present disclosure. It should be understood that it is included in the technical idea.

Meanwhile, a notification message according to an aspect of the present disclosure may be provided for various purposes. For example, the notification message may include an injury risk warning message to the target entity. If the degree of difference between the estimated load index and the response load index satisfies the criterion, the corresponding target entity may be interpreted as showing a prognostic symptom of injury. Accordingly, by including the injury risk warning message in the notification message, the administrator can check the notification message and take precautionary measures to prevent injury to the corresponding target entity.

According to another aspect, the notification message may include an abnormal condition warning message for the target entity. Satisfaction of the difference criterion may be interpreted as the fact that the target entity is not in a normal condition state. Therefore, by including the abnormal condition warning message in the notification message, the administrator can use it to determine the roster for the next game session or initiate condition recovery measures.

According to another aspect, the notification message may include a coach interview request message for the target entity. Abnormal symptoms of the target entity may be due to psychological problems such as family or personal life issues as well as physical problems. Accordingly, by including the coach interview request message in the notification message, the manager can initiate a coach interview with the corresponding entity.

According to another aspect, the notification message may include a type change recommendation message for an artificial neural network. As described above, according to one aspect of the present disclosure, target entities may be classified into a plurality of groups according to a predetermined criterion, and each artificial neural network may be trained using a training data set corresponding to each classification. Therefore, if the criterion for the difference between the estimated load index and the response load index for a target entity is met, the notification message includes a type change recommendation message for the artificial neural network, so that the system administrator can determine that the appropriate artificial neural network for that entity is It is applied to check whether the estimated load index is calculated, and the applied artificial neural network can be changed according to the case.

Meanwhile, in one aspect of the present disclosure, a target entity to be monitored or an anomaly detection may be at least one sports team or one sports player.

According to one aspect, when the target entity is at least one sports team, each of the estimated load index and the response load index may be a representative index determined based on indices for each of at least some athletes belonging to the sports team. For example, it may be an average value of indices for each of a plurality of athletes, but is not limited thereto, and any criterion for representing the index for each of a plurality of athletes may be utilized. According to one aspect, by monitoring the difference between the estimated load index and the response load index of the sports team and detecting an anomaly, it is possible to determine whether there is a problem related to the management of the sports team, for example It can be checked for various problems such as diet, facilities, and problems outside of training.

According to one aspect, when the target entity is a single sports player, a problem of the sports player may be reviewed. The risk of injury of the sports player may be detected at an early stage, and whether or not psychologically affecting matters such as family problems or personal life issues exist in the sports player may be reviewed.

By performing such monitoring on the target entity, an effect of preventing injury of the target entity or improving the possibility of securing an optimal condition can be expected. In addition, in the management of clubs in professional sports leagues that record astronomical annual salaries, it can help improve efficiency in terms of costs invested in target entities.

avatar

Hereinafter, exemplary implementations of a process for providing exercise load information according to an aspect of the present disclosure will be described. However, it should be noted that the technical spirit of the present description is not limited by the examples described.

40 illustrates a schematic structure of an artificial neural network for implementing a method for providing exercise load information according to an embodiment of the present disclosure. In FIG. 40 , the dotted line represents the information flow of the training phase and the solid line represents the information flow of the inference phase.

The activity information may represent information reflecting the physical activity of a sports player performing an exercise, such as kinematic information related to the position and/or movement of the sports player and physical condition information related to the player's body, related to the player's activity. may include information about any characteristics that change due to

“qRPE” may indicate an exercise load index obtained from a player through an inquiry procedure following the end of an exercise session. For example, "qRPE" may be understood as a response load index.

“pRPE” may indicate a value of RPE predicted by an artificial neural network model. For example, "pRPE" can be understood as an estimated load index.

As shown in FIG. 40 , in the training phase, activity information 3710 of a specific entity, for example, can be labeled as qRPE 3750 . That is, the activity information 3710 of a specific entity labeled with the value of qRPE 3750 can be used as a learning data set for training the artificial neural network 3720. Here, the artificial neural network may be trained so that the value of pRPE (3730) output by the artificial neural network is close to the value of qRPE (3750).

Referring back to FIG. 40 , in the inference step, activity information 3710 in a target exercise session of a target entity for which exercise load information is to be determined may be input to the artificial neural network 3720 . The artificial neural network outputs a predetermined value of pRPE 3730, and based on this pRPE 3730, an estimated load index 3740 is determined to estimate an exercise load for exercise load information for a target exercise session of a target entity. Index 3740 can be provided.

41 illustrates an embodiment of providing exercise load information using kinematic information. As shown in FIG. 41 , first, a wearable device corresponding to a target entity for which exercise load information is to be determined may be mounted (4110). For the workout session 4120 , data about the position or orientation of an entity based on the wearable device can be tracked 4130 . Kinematic information 4140 can then be determined using the tracking data.

Meanwhile, in the training phase, kinematic information may be labeled with a response load index (eg, qRPE) value 4150 received by querying an entity. The artificial neural network 4170 may be trained so that the value of the estimated load index (eg, pRPE) 4160 output when the kinematic information 4140 is input is close to the value of the response load index 4150.

Meanwhile, in the utilization step, when the kinematic information 4140 is input to the artificial neural network 4170, the output estimated load index 4180 value may be provided as exercise load information. According to one aspect, the estimated load index 4180 output and the response load index 4150 may be compared to perform monitoring and early risk detection of a target entity based on whether a difference between the two meets a predetermined criterion. there is.

Also, according to one aspect, the pRPE value may not reflect individual characteristics or daily conditions of an entity. For example, assuming that the physical level of the target entity is similar to that of the entities corresponding to the learning data set used to train the artificial neural network, information about the difference between qRPE and pRPE is the target entity It may be used as a daily conditioning factor of

42 is an exemplary diagram of an artificial neural network structure considering time information. As shown in FIG. 42 , for example, like a sequence of kinematic information, the target data set may include a sequence of feature values corresponding to each of a plurality of time points or time units included in the target exercise session. The artificial neural network used to implement the exercise load information providing method according to one aspect of the present disclosure divides the sequences included in the target data set into two or more groups in chronological order, for example, and generates a convolutional neural network for each group ( Embedding for Convolution Neural Network (CNN) can be performed. As shown in FIG. 42 , for example, a sequence 4210 of kinematic information can be divided into a first group 4211, a second group 4213, and a third group 4215 according to time order. A first group 4211, a second group 4213, and a third group 4215 for the sequence of kinematic information can be embedded, respectively, and the embedding result 4230 is applied to a recurrent neural network (RNN). It can be input so that the exercise load index (eg, RPE having a value of 0 to 10) 4250 outputs it. Here, the time interval T may be a predetermined time interval such as 5 minutes or 10 minutes, for example.

According to one aspect of the present disclosure, when a target entity performs an exercise session, an estimated value of an exercise load level up to a current point in time may be estimated in real time. For example, a real-time estimated load index for a current point in time may be determined based on a sequence of kinematic information up to the current point in time. Therefore, based on the real-time estimated load index, it may be used as a reference for the necessity of replacing players in a game session, or it may be decided to stop performing an exercise session to prevent injury.

43 is an exemplary view of a method of providing real-time exercise load information according to an aspect of the present disclosure. 44 is an exemplary view of kinematic data up to time point N in FIG. 43 , and FIG. 45 is an example view of kinematic data up to time point M in FIG. 43 .

As shown in FIG. 43 , first, a corresponding wearable device may be mounted on a target entity for which exercise load information is to be determined (4310). For the exercise session 4320 , data about the location or orientation of an entity based on the wearable device can be tracked 4330 .

Next, kinematic information 4340 up to time point N may be determined using tracking data up to time point N (T ^N ). As shown in FIG. 44 , kinematic information to time N 4340 can include a sequence of kinematic information of the target entity from the start of the exercise session to time N. As shown in FIG. 43 , if kinematic information 4340 of the target entity up to time N is input to the artificial neural network 4350, a real-time estimated load index 4360 for the target entity at time N can be determined. Here, since the data input to the artificial neural network may be configured to have a predetermined length, for example, as shown in FIG. 44, zero padding follows or precedes the kinematic information sequence of the target entity up to N can be performed.

Referring back to FIG. 43 , kinematic information 4370 up to time point M may be determined using tracking data up to time point M (T ^M ) following time point N. As shown in FIG. 45 , kinematic information to time M 4370 can include a sequence of kinematic information of the target entity from the start of the exercise session to time M. As shown in FIG. 43 , if kinematic information 4370 of the target entity up to time M is input to the artificial neural network 4380, a real-time estimated load index 4390 for the target entity at time M can be determined. Here, since data input to the artificial neural network may be configured to have a predetermined length, for example, as shown in FIG. can be performed.

In this way, a real-time estimated load index for time N or M, which is a specific time point of the exercise session, may be determined. Based on the real-time load index for a specific point in time, monitoring and critical load warning procedures for the target entity may be performed. 46 illustrates a critical load warning procedure according to real-time exercise load measurement. As shown in FIG. 46 , after the start of the exercise session, a real-time estimated load index may be calculated at predetermined time intervals to monitor whether or not the threshold level has elapsed. If monitoring is performed at time N, a warning message may not be delivered because the real-time estimated load index at time N does not exceed the threshold level. On the other hand, if monitoring is performed at time M, since the real-time estimated load index at time M has exceeded the threshold level, a warning message can be delivered. The monitoring interval can be adjusted by various factors such as system performance or administrator settings. Through the monitoring procedure as described above, it is possible to provide a reference on whether a player is replaced during a game, and to achieve an injury prevention effect by terminating an exercise session for a player at risk of injury.

The method according to the present invention described above may be implemented as computer readable codes on a computer readable recording medium. Computer-readable recording media includes all types of recording media in which data that can be decoded by a computer system is stored. For example, there may be read only memory (ROM), random access memory (RAM), a magnetic tape, a magnetic disk, a flash memory, an optical data storage device, and the like. In addition, the computer-readable recording medium may be distributed in computer systems connected through a computer communication network, and stored and executed as readable codes in a distributed manner.

Although the above has been described with reference to the drawings and examples, it does not mean that the scope of protection of the present invention is limited by the drawings or examples, and those skilled in the art will understand the scope of the present invention described in the claims below. It will be understood that various modifications and changes may be made to the present invention without departing from the spirit and scope.

Specifically, the described features may be implemented within digital electronic circuitry, or within computer hardware, firmware, or combinations thereof. Features may be implemented in a computer program product embodied within storage, eg, in a machine-readable storage device, for execution by a programmable processor. And features can be performed by a programmable processor executing a program of instructions to perform the functions of the described embodiments by operating on input data and generating output. The described features include at least one programmable processor, at least one input device, and at least one output device coupled to receive data and instructions from and to transmit data and instructions to the data storage system. It can be executed within one or more computer programs that can be executed on a programmable system including. A computer program contains a set of instructions that can be used directly or indirectly within a computer to perform a particular action for a given result. A computer program is written in any programming language, including compiled or interpreted languages, and contained as modules, components, subroutines, or other units suitable for use in other computer environments, or as stand-alone programs. can be used in any form.

Suitable processors for execution of a program of instructions include, for example, both general and special purpose microprocessors, and either a single processor or multiple processors in a computer of another type. Also, storage devices suitable for embodying computer program instructions and data embodying the described features include, for example, semiconductor memory devices such as EPROM, EEPROM, and flash memory devices, magnetic memory devices such as internal hard disks and removable disks. devices, magneto-optical disks and all forms of non-volatile memory including CD-ROM and DVD-ROM disks. The processor and memory may be integrated within or added by ASICs (application-specific integrated circuits).

Although the present invention described above has been described based on a series of functional blocks, it is not limited by the above-described embodiments and the accompanying drawings, and various substitutions, modifications, and changes are made within the scope of the technical spirit of the present invention. That this is possible will be apparent to those skilled in the art.

Combinations of the above-described embodiments are not limited to the above-described embodiments, and various types of combinations may be provided as well as the above-described embodiments according to implementation and/or needs.

In the foregoing embodiments, the methods are described on the basis of a flow chart as a series of steps or blocks, but the present invention is not limited to the order of steps, and some steps may occur in a different order or concurrently with other steps as described above. there is. In addition, those skilled in the art will understand that the steps shown in the flow chart are not exclusive, that other steps may be included, or that one or more steps of the flow chart may be deleted without affecting the scope of the present invention. You will understand.

The foregoing embodiment includes examples of various aspects. It is not possible to describe all possible combinations to represent the various aspects, but those skilled in the art will recognize that other combinations are possible. Accordingly, it is intended that the present invention cover all other substitutions, modifications and variations falling within the scope of the following claims.

Claims (15)

As a method of providing exercise load information for a target entity,
A target data set for a target entity during a target workout session comprising a plurality of time units, the target data set comprising a sequence of kinematic information of the target entity for each of the plurality of time units and the obtaining - including condition information reflecting characteristics associated with the target exercise session; and
determining an estimated load index reflecting a degree of exercise load of the target exercise session for the target entity, using an artificial neural network, from the target data set; including,
How to provide exercise load information.
According to claim 1,
Determining the estimated load index,
a plurality of training data sets comprising a first training data set, the first training data set comprising a sequence of kinematic information of a first entity during a first workout session and state information for the first workout session; preparing the artificial neural network trained based on - labeled with a score collected from the first entity for an exercise load of the first exercise session;
inputting the target data set to an input layer of the artificial neural network; and
obtaining the estimated load index based on a result value of an output layer of the artificial neural network; including,
How to provide exercise load information.
According to claim 1,
The status information is
Including daily readiness response information collected for the target exercise session from the target entity,
How to provide exercise load information.
According to claim 3,
The daily readiness response information,
sleep time evaluation value;
sleep quality evaluation values;
fatigue level evaluation value;
pain level (soreness level) evaluation value;
stress level evaluation value; or
mood assessment value; including at least one of
How to provide exercise load information.
According to claim 1,
The status information is
Including exercise type information reflecting information on whether the target exercise session is a match session or a training session,
How to provide exercise load information.
According to claim 1,
The status information is
Including an importance evaluation value according to the type of the target exercise session,
How to provide exercise load information.
According to claim 1,
The status information is
containing information about the length of time elapsed from the previous game session to the target workout session;
How to provide exercise load information.
According to claim 1,
The status information is
Including information on the length of time that has elapsed from the start date of the sports season including the target exercise session to the target exercise session,
How to provide exercise load information.
According to claim 1,
The status information is
Including information about the physical condition of the target entity,
How to provide exercise load information.
According to claim 9,
Information about the physical condition,
age;
key;
weight;
fitness rating;
muscle strength evaluation value;
maximum heart rate; or
maximal oxygen uptake; including at least one of
How to provide exercise load information.
According to claim 1,
The status information is
Including environmental information reflecting at least one of weather information, temperature information, humidity information, or season information for the target exercise session,
How to provide exercise load information.
According to claim 1,
The status information is
Comprising biological measurement information of the target entity during the target exercise session,
How to provide exercise load information.
According to claim 12,
The biological measurement information,
heart rate information;
information about the ratio of heart rate to maximum heart rate;
body temperature information;
oxygen uptake information; or
information on the ratio of oxygen uptake to maximal oxygen uptake; including at least one of
How to provide exercise load information.
A device for providing exercise load information for a target entity, the device including a processor and a memory, the processor comprising:
A target data set for a target entity during a target workout session comprising a plurality of time units, the target data set comprising a sequence of kinematic information of the target entity for each of the plurality of time units and the obtains - including condition information reflecting characteristics associated with the target workout session; and
determining an estimated load index reflecting an exercise load degree of the target exercise session for the target entity, using an artificial neural network, from the target data set; configured to
A device that provides exercise load information.
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:
A target data set for a target entity during a target workout session comprising a plurality of time units, the target data set comprising a sequence of kinematic information of the target entity for each of the plurality of time units and the obtains - including condition information reflecting characteristics associated with the target workout session; and
determining an estimated load index reflecting an exercise load degree of the target exercise session for the target entity, using an artificial neural network, from the target data set; configured to make
A computer readable storage medium.

Images