CN112802067B - Multi-target tracking method and system based on graph network - Google Patents

Multi-target tracking method and system based on graph network Download PDF

Info

Publication number
CN112802067B
CN112802067B CN202110104067.XA CN202110104067A CN112802067B CN 112802067 B CN112802067 B CN 112802067B CN 202110104067 A CN202110104067 A CN 202110104067A CN 112802067 B CN112802067 B CN 112802067B
Authority
CN
China
Prior art keywords
tracking
preset
state
target
graph
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110104067.XA
Other languages
Chinese (zh)
Other versions
CN112802067A (en
Inventor
杨培春
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Puhui Zhilian Technology Co ltd
Original Assignee
Shenzhen Puhui Zhilian Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Puhui Zhilian Technology Co ltd filed Critical Shenzhen Puhui Zhilian Technology Co ltd
Priority to CN202110104067.XA priority Critical patent/CN112802067B/en
Publication of CN112802067A publication Critical patent/CN112802067A/en
Application granted granted Critical
Publication of CN112802067B publication Critical patent/CN112802067B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/292—Multi-camera tracking
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10016—Video; Image sequence
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20212—Image combination
    • G06T2207/20221—Image fusion; Image merging
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30221—Sports video; Sports image
    • G06T2207/30224—Ball; Puck
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30241—Trajectory

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Image Analysis (AREA)

Abstract

The invention provides a multi-target tracking method and a system based on a graph network, wherein the method comprises the following steps: determining a plurality of tracking targets according to the operation instruction of the user; tracking each tracking target according to a preset tracking method; in the tracking process, if any tracking target fails to track, acquiring a first motion track generated by the tracking target within a first time period preset before the tracking failure; and based on a second motion track generated again by the tracked target in a second time period preset after the tracking failure of the graph network acquisition, correlating the first motion track with the second motion track to obtain a complete motion track. The multi-target tracking method and system based on the graph network greatly improves the integrity of tracking the track of the billiard ball.

Description

Multi-target tracking method and system based on graph network
Technical Field
The invention relates to the technical field of video monitoring, in particular to a multi-target tracking method and system based on a graph network.
Background
At present, an image recognition technology is mostly adopted when each billiard ball on a billiard table is tracked, but in practical application, when the billiard ball is hit and moves at a high speed, the picture collected by a common camera is fuzzy due to the fact that the speed is high, tracking loss is easy to occur, in addition, the billiard ball is frequently shielded by a user body and a billiard ball rod, tracking loss is also caused, and therefore when a tracking result is displayed, the multi-section movement track of the billiard ball cannot be quickly associated, and the integrity of tracking the billiard ball is affected.
Disclosure of Invention
The invention aims to provide a multi-target tracking method and a system based on a graph network, which are used for acquiring a first motion track generated before a billiard ball on a billiard table fails to track, determining a second motion track generated after the billiard ball fails to track based on the graph network, and correlating the first motion track with the second motion track to be used as a complete motion track of the billiard ball, so that the problems that a camera acquires an image in a fuzzy way, a user body shields the billiard ball and a ball rod shields the billiard ball, and the correlation of multiple segments of motion tracks of the billiard ball cannot be performed quickly are avoided, and the integrity of tracking the billiard ball is greatly improved.
The embodiment of the invention provides a multi-target tracking method based on a graph network, which comprises the following steps:
determining a plurality of tracking targets according to the operation instruction of the user;
tracking each tracking target according to a preset tracking method;
in the tracking process, if any tracking target fails to track, acquiring a first motion track generated by the tracking target within a first time period preset before the tracking failure;
and based on a second motion track generated again by the tracked target in a second time period preset after the tracking failure of the graph network acquisition, correlating the first motion track with the second motion track to obtain a complete motion track.
Preferably, the graph network includes: the device comprises a feature extraction module and a feature matching module.
Preferably, the second motion track generated again by the tracking target in a second time period preset after the tracking failure is acquired based on the graph network specifically includes:
acquiring a first image corresponding to a first time preset before tracking failure and a second image corresponding to a second time preset after tracking failure;
extracting a first feature of a tracking target on a first image through a feature extraction module;
extracting second features of each tracking target on the second image through a feature extraction module, and combining the second features into a second feature set;
selecting any one of the second features in the second feature set as a target feature;
and judging whether the first feature is matched with the target feature or not through a feature matching module, and if so, acquiring a second motion track generated by a tracking target corresponding to the target feature in a second time period.
Preferably, the graph network-based multi-target tracking method further comprises:
acquiring state parameters of a graph network, processing the state parameters to acquire a graph network state score, acquiring tracking parameters of each tracking target, processing the tracking parameters to acquire a table state index, judging whether the graph network state score is matched with the table state index, and executing a corresponding preset strategy if the graph network state score is not matched with the table state index;
Wherein the status parameters include: the state value and the state prediction value of each node in the graph network;
processing the state parameters to obtain a map network state score, which specifically comprises the following steps:
calculating a graph network state score based on the state values and the state predictors:
where α is the graph network state score, n is the total number of nodes in the graph network, x i For the state value of the ith node in the graph network, y i State predictive value, x, for the i-th node in the graph network i，0 A preset state value threshold corresponding to an ith node in the graph network;
the tracking parameters include: real-time position and real-time speed of each tracking target;
processing the tracking parameters to obtain a table state index, which specifically comprises the following steps:
calculating a table state index based on the real-time position and the real-time speed:
wherein f is a table state index, v t For the t-th real-time speed of the tracking target, v 0 For a preset speed threshold, s t For the t-th tracking target real-time position, s e For the e-th real-time position of the tracking target, L is a preset distance function for obtaining the distance between the two positions, L 0 Z is the total number of the tracking targets and is a preset distance threshold;
judging whether the network state score of the graph is matched with the state index of the table or not, specifically comprising the following steps:
Acquiring a first interval corresponding to the network state score of the graph in a preset database of the state indexes of the suitable table;
acquiring a second interval corresponding to the table state index in a preset network state scoring interval database of the suitable graph;
calculating a matching index based on the first interval and the second interval:
wherein match is a matching index, f is a table state index, α is a graph network score, u 1 Is the upper limit value of the first interval, d 1 As the lower limit value of the first interval, u 2 Is the upper limit value of the second interval, d 2 Is the lower limit value of the second interval,and->The weight value is preset;
and when the matching index is smaller than or equal to a preset matching index threshold, determining that the network state score of the graph is not matched with the table state index.
Preferably, the graph network-based multi-target tracking method further comprises:
when any tracking target fails to track, if the time of the tracking failure is greater than or equal to a preset first time threshold, simulating a simulated motion track of the tracking target in time;
the method for simulating the simulated motion trail of the tracking target in time specifically comprises the following steps:
taking the tracking target as a pre-simulation target;
acquiring the instantaneous position, the instantaneous speed and the instantaneous direction of a pre-simulated target at a third moment preset before tracking failure;
Simulating a simulated motion trail of a pre-simulated target based on the instantaneous position, the instantaneous speed and the instantaneous direction;
the method further comprises the steps of:
if the tracking failure time is greater than or equal to a preset second time threshold, a corresponding preset reminding picture is projected on the tabletop of the billiard table through a projection device, and reminding is carried out on the user;
the method comprises the steps of projecting corresponding preset reminding pictures on a tabletop of a billiard table through a projection device, and specifically comprises the following steps:
acquiring a preset initial graph;
projecting the initial graph onto a desktop;
during the projection process, the side length of the initial graph is adjusted based on the time of tracking failure:
wherein L is the side length of the initial graph after determination, L 0 For the preset initial value of the side length, T is tracking failureTime, T of (1) 0 Is a second time threshold.
The embodiment of the invention provides a multi-target tracking system based on a graph network, which comprises the following components:
the determining module is used for determining a plurality of tracking targets according to the operation instruction of the user;
the tracking module is used for tracking each tracking target according to a preset tracking method;
the acquisition module is used for acquiring a first motion track generated by any tracking target in a preset first time period before the tracking failure if the tracking of the tracking target fails in the tracking process;
And the determining and associating module is used for associating the first motion trail and the second motion trail as complete motion trail based on a second motion trail which is generated again by the tracked target in a second time period and is preset after the tracking failure of the graph network acquisition.
Preferably, the graph network includes: the device comprises a feature extraction module and a feature matching module.
Preferably, the determining and associating module performs operations including:
acquiring a first image corresponding to a first time preset before tracking failure and a second image corresponding to a second time preset after tracking failure;
extracting a first feature of a tracking target on a first image through a feature extraction module;
extracting second features of each tracking target on the second image through a feature extraction module, and combining the second features into a second feature set;
selecting any one of the second features in the second feature set as a target feature;
and judging whether the first feature is matched with the target feature or not through a feature matching module, and if so, acquiring a second motion track generated by a tracking target corresponding to the target feature in a second time period.
Preferably, the graph network-based multi-target tracking system further comprises:
the judging and executing module is used for obtaining state parameters of the graph network, processing the state parameters to obtain graph network state scores, obtaining tracking parameters of each tracking target, processing the tracking parameters to obtain a table state index, judging whether the graph network state scores are matched with the table state index, and executing corresponding preset strategies if the graph network state scores are not matched with the table state index;
Wherein the status parameters include: the state value and the state prediction value of each node in the graph network, and the tracking parameters comprise: real-time position and real-time speed of each tracking target;
the judging and executing module executes the following operations:
calculating a graph network state score based on the state values and the state predictors:
where α is the graph network state score, n is the total number of nodes in the graph network, x i For the state value of the ith node in the graph network, y i State predictive value, x, for the i-th node in the graph network i，0 A preset state value threshold corresponding to an ith node in the graph network;
calculating a table state index based on the real-time position and the real-time speed:
wherein f is a table state index, v t For the t-th real-time speed of the tracking target, v 0 For a preset speed threshold, s t For the t-th tracking target real-time position, s e For the e-th real-time position of the tracking target, L is a preset distance function for obtaining the distance between the two positions, L 0 Z is the total number of the tracking targets and is a preset distance threshold;
the judging and executing module judges whether the network state score of the graph is matched with the state index of the table by adopting the following preset method, and specifically comprises the following steps:
Acquiring a first interval corresponding to the network state score of the graph in a preset database of the state indexes of the suitable table;
acquiring a second interval corresponding to the table state index in a preset network state scoring interval database of the suitable graph;
calculating a matching index based on the first interval and the second interval:
wherein match is a matching index, f is a table state index, α is a graph network score, u 1 Is the upper limit value of the first interval, d 1 As the lower limit value of the first interval, u 2 Is the upper limit value of the second interval, d 2 Is the lower limit value of the second interval,and->The weight value is preset;
and when the matching index is smaller than or equal to a preset matching index threshold, determining that the network state score of the graph is not matched with the table state index.
Preferably, the graph network-based multi-target tracking system further comprises:
the simulation module is used for simulating the simulated motion trail of any tracking target in time when the tracking failure time is greater than or equal to a preset first time threshold value;
wherein the simulation module performs operations comprising:
taking the tracking target as a pre-simulation target;
acquiring the instantaneous position, the instantaneous speed and the instantaneous direction of a pre-simulated target at a third moment preset before tracking failure;
Simulating a simulated motion trail of a pre-simulated target based on the instantaneous position, the instantaneous speed and the instantaneous direction;
the system further comprises:
the projection reminding module is used for projecting a corresponding preset reminding picture on the tabletop of the billiard table through the projection device to remind a user if the tracking failure time is greater than or equal to a preset second time threshold;
the projection reminding module performs the following operations:
acquiring a preset initial graph;
projecting the initial graph onto a desktop;
during the projection process, the side length of the initial graph is adjusted based on the time of tracking failure:
wherein L is the side length of the initial graph after determination, L 0 For the preset initial value of the side length, T is the time of tracking failure, T 0 Is a second time threshold.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.
The technical scheme of the invention is further described in detail through the drawings and the embodiments.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:
FIG. 1 is a flow chart of a multi-objective tracking method based on graph network in an embodiment of the invention;
fig. 2 is a schematic diagram of a multi-target tracking system based on a graph network according to an embodiment of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described below with reference to the accompanying drawings, it being understood that the preferred embodiments described herein are for illustration and explanation of the present invention only, and are not intended to limit the present invention.
The embodiment of the invention provides a multi-target tracking method based on a graph network, which is shown in fig. 1 and comprises the following steps:
s1, determining a plurality of tracking targets according to operation instructions of a user;
s2, tracking each tracking target according to a preset tracking method;
s3, in the tracking process, if any tracking target fails to track, acquiring a first motion track generated by the tracking target within a first time period preset before the tracking failure;
s4, acquiring a second motion track generated again by the tracked target in a second time period preset after the tracking failure based on the graph network, and associating the first motion track with the second motion track to serve as a complete motion track.
The working principle of the technical scheme is as follows:
a camera is arranged at a position at least 1.6 meters above the billiard table; a user may operate (e.g., select a plurality of billiards to be tracked) by using a smart terminal (e.g., a smart phone, tablet, computer, etc.); the preset method specifically comprises the following steps: a target detection algorithm; when tracking fails (for example, billiards are shielded by a club or a body of a user and image blur detection acquired by a camera fails), acquiring a first motion track generated by billiards in a preset first time period (for example, 10 seconds before the tracking fails), acquiring a second motion track generated by billiards in a preset second time period (5 seconds after the tracking fails) based on a graph network, and associating the first motion track and the second motion track to form a complete motion track of the billiards; the billiards can be tracked to judge whether the billiards fall into bags, and the bag falling automatically stops tracking.
The beneficial effects of the technical scheme are as follows: according to the embodiment of the invention, when the billiard ball on the billiard table fails to track, the first movement track generated before the billiard ball fails to track is acquired, the second movement track generated after the billiard ball fails to track is determined based on the graph network, and the first movement track and the second movement track are correlated to be used as the complete movement track of the billiard ball, so that the problems that the image collected by a camera is blurred, the billiard ball is blocked by a user body, and the billiard ball is blocked by a ball rod, the multi-section movement track of the billiard ball cannot be correlated quickly, and the integrity of tracking the billiard ball is greatly improved.
The embodiment of the invention provides a multi-target tracking method based on a graph network, which comprises the following steps: the device comprises a feature extraction module and a feature matching module.
The working principle of the technical scheme is as follows: the graph network has the functions of feature extraction and feature matching, and is not described in detail in the prior art.
The beneficial effects of the technical scheme are as follows: the graph network of the embodiment of the invention comprises a feature extraction module and a feature matching module.
The embodiment of the invention provides a multi-target tracking method based on a graph network, which is based on a second motion track generated again by a tracking target in a second time period preset after the acquisition of tracking failure of the graph network, and specifically comprises the following steps:
acquiring a first image corresponding to a first time preset before tracking failure and a second image corresponding to a second time preset after tracking failure;
extracting a first feature of a tracking target on a first image through a feature extraction module;
extracting second features of each tracking target on the second image through a feature extraction module, and combining the second features into a second feature set;
selecting any one of the second features in the second feature set as a target feature;
and judging whether the first feature is matched with the target feature or not through a feature matching module, and if so, acquiring a second motion track generated by a tracking target corresponding to the target feature in a second time period.
The working principle of the technical scheme is as follows:
the preset first time is specifically: tracking the last moment before failure; the preset second moment is specifically: tracking the corresponding moments when the number of billiards (including clear recognition and fuzzy recognition) on the billiards table after failure meets a reasonable numerical value; for example: after a billiard ball is opened for a period of time, 10 billiards are left on the tabletop, after a certain billiard ball fails to track, no billiard ball falls into a bag during the period, and when 10 billiards appear on the tabletop again, the billiards are automatically tracked; the first and second features are specifically: appearance characteristics, spatial relationship characteristics and time relationship characteristics of billiards on the corresponding images.
The beneficial effects of the technical scheme are as follows: according to the embodiment of the invention, the second motion trail generated again by the tracking target (namely the billiard ball) in the preset second time period after the tracking failure is determined through the graph network, and the effect is accurate.
The embodiment of the invention provides a multi-target tracking method based on a graph network, which further comprises the following steps:
acquiring state parameters of a graph network, processing the state parameters to acquire a graph network state score, acquiring tracking parameters of each tracking target, processing the tracking parameters to acquire a table state index, judging whether the graph network state score is matched with the table state index, and executing a corresponding preset strategy if the graph network state score is not matched with the table state index;
Wherein the status parameters include: the state value and the state prediction value of each node in the graph network;
processing the state parameters to obtain a map network state score, which specifically comprises the following steps:
calculating a graph network state score based on the state values and the state predictors:
where α is the graph network state score, n is the total number of nodes in the graph network, x i For the state value of the ith node in the graph network, y i State predictive value, x, for the i-th node in the graph network i，0 A preset state value threshold corresponding to an ith node in the graph network;
the tracking parameters include: real-time position and real-time speed of each tracking target;
processing the tracking parameters to obtain a table state index, which specifically comprises the following steps:
calculating a table state index based on the real-time position and the real-time speed:
wherein f is a table state index, v t For the t-th real-time speed of the tracking target, v 0 For a preset speed threshold, s t For the t-th tracking target real-time position, s e For the e-th real-time position of the tracking target, L is a preset distance function for obtaining the distance between the two positions, L 0 Z is the total number of the tracking targets and is a preset distance threshold;
judging whether the network state score of the graph is matched with the state index of the table or not, specifically comprising the following steps:
Acquiring a first interval corresponding to the network state score of the graph in a preset database of the state indexes of the suitable table;
acquiring a second interval corresponding to the table state index in a preset network state scoring interval database of the suitable graph;
calculating a matching index based on the first interval and the second interval:
wherein match is a matching index, f is a table state index, α is a graph network score, u 1 Is the upper limit value of the first interval, d 1 As the lower limit value of the first interval, u 2 Is the upper limit value of the second interval, d 2 Is the lower limit value of the second interval,and->The weight value is preset;
and when the matching index is smaller than or equal to a preset matching index threshold, determining that the network state score of the graph is not matched with the table state index.
The working principle of the technical scheme is as follows:
calculating a graph network state score based on a state value (such as a load flow value) and a state prediction value (such as a value for predicting a current state value after a large amount of historical operation data is learned through a pre-trained learning model) of each node in the graph network, wherein the lower the score is, the higher the load of the graph network is, and the worse the state is; the speed of billiards on a billiard table is high, when a plurality of billiards are close to the billiards, the probability of collision of the billiards is high, the billiards need to be ready for tracking, the situation of tracking loss is high due to the fact that the speed of each billiard is high at the moment of collision, and the total load of a graph network is high, therefore, the state index of the billiard table is calculated based on the real-time position and the real-time speed of each billiard, and the higher the state index is, the higher the probability of collision of the billiards is; when the state score of the graph network is not matched with the state index of the table, the graph network is insufficient in tracking loss processing performance, and a preset strategy (for example, a standby node is started) is required to be executed; the preset database suitable for the table state index interval is specifically: the table state index interval corresponding to each pre-stored network state score of each graph is, for example: if the network state score of the graph is 85, the corresponding appropriate table state index interval is [80,90]; the preset network state scoring interval database of the suitable graph is specifically: a network state scoring interval of a suitable map corresponding to each table state index is pre-stored, for example: if the table state index is 65, the corresponding network state scoring interval of the suitable graph is [60,70]; the preset matching index threshold value is specifically: for example 30.
The beneficial effects of the technical scheme are as follows: according to the embodiment of the invention, the running state of the graph network and the movement state of a plurality of billiards on the billiards table are respectively evaluated by calculating the graph network scores and the billiards table state indexes, when the graph network scores and the billiards table state indexes are not matched, the defect of insufficient processing performance of the graph network on tracking loss is illustrated, a preset strategy is required to be executed immediately, the fact that the speed of each billiard is higher at the moment of collision when a plurality of billiards are collided is ensured, and the processing performance of the graph network is greatly improved when the tracking loss is increased; and calculating a matching index based on the first interval and the second interval, and judging that the current network state score is not matched with the table state index when the matching index is smaller than or equal to a preset matching index threshold value, wherein the judgment is accurate.
The embodiment of the invention provides a multi-target tracking method based on a graph network, which further comprises the following steps:
when any tracking target fails to track, if the time of the tracking failure is greater than or equal to a preset first time threshold, simulating a simulated motion track of the tracking target in time;
the method for simulating the simulated motion trail of the tracking target in time specifically comprises the following steps:
Taking the tracking target as a pre-simulation target;
acquiring the instantaneous position, the instantaneous speed and the instantaneous direction of a pre-simulated target at a third moment preset before tracking failure;
simulating a simulated motion trail of a pre-simulated target based on the instantaneous position, the instantaneous speed and the instantaneous direction;
the method further comprises the steps of:
if the tracking failure time is greater than or equal to a preset second time threshold, a corresponding preset reminding picture is projected on the tabletop of the billiard table through a projection device, and reminding is carried out on the user;
the method comprises the steps of projecting corresponding preset reminding pictures on a tabletop of a billiard table through a projection device, and specifically comprises the following steps:
acquiring a preset initial graph;
projecting the initial graph onto a desktop;
during the projection process, the side length of the initial graph is adjusted based on the time of tracking failure:
wherein L is the side length of the initial graph after determination, L 0 For the preset initial value of the side length, T is the time of tracking failure, T 0 Is a second time threshold.
The working principle of the technical scheme is as follows:
if the time of the failure tracking is greater than or equal to the preset first time threshold (for example, 5 seconds), if the billiard ball still moves in the period of time, the movement track of the billiard ball in the period of time is longer (two tracks can be directly connected in a shorter period of time), in order to display the whole track, a simulated movement track can be simulated based on the instant position, the instant speed and the instant direction of the billiard ball at the preset third moment (last moment before the failure tracking), and the simulated movement track can be fused with the first movement track and the second movement track to be used as the whole movement track of the billiard ball; if the tracking failure time is greater than or equal to a preset second time threshold (for example, 15 seconds), the user is informed that the unintentional shielding time is longer (for example, other users rely on the billiard table for a long time and the head and the like of the other users shield billiard balls), and the billiard ball is required to be reminded; the projection state is specifically: a projector; the preset initial graph is specifically: the frame is yellow, and the character of 'please remove the club/body and thank you' is arranged in the frame; if the tracking failure time is longer, the side length of the regular polygon can be adjusted, namely, the pattern range is enlarged, and a user is reminded to timely remove the club and the like.
The beneficial effects of the technical scheme are as follows: according to the embodiment of the invention, when the tracking failure time is greater than or equal to the first time threshold, the simulated motion track of the corresponding billiard ball in the period of the tracking failure is simulated, so that the track tracking integrity is improved.
The embodiment of the invention provides a multi-target tracking system based on a graph network, which is shown in fig. 2 and comprises the following components:
a determining module 1, configured to determine a plurality of tracking targets according to an operation instruction of a user;
the tracking module 2 is used for tracking each tracking target according to a preset tracking method;
the acquisition module 3 is used for acquiring a first motion track generated by any tracking target in a first time period preset before the tracking failure if the tracking of the tracking target fails in the tracking process;
and the determining and associating module 4 is used for associating the first motion trail and the second motion trail as complete motion trail based on a second motion trail which is generated again by the tracked target in a second time period preset after the tracking failure of the graph network acquisition.
The working principle of the technical scheme is as follows:
a camera is arranged at a position at least 1.6 meters above the billiard table; a user may operate (e.g., select a plurality of billiards to be tracked) by using a smart terminal (e.g., a smart phone, tablet, computer, etc.); the preset method specifically comprises the following steps: a target detection algorithm; when tracking fails (for example, billiards are shielded by a club or a body of a user and image blur detection acquired by a camera fails), acquiring a first motion track generated by billiards in a preset first time period (for example, 10 seconds before the tracking fails), acquiring a second motion track generated by billiards in a preset second time period (5 seconds after the tracking fails) based on a graph network, and associating the first motion track and the second motion track to form a complete motion track of the billiards; the billiards can be tracked to judge whether the billiards fall into bags, and the bag falling automatically stops tracking.
The beneficial effects of the technical scheme are as follows: according to the embodiment of the invention, when the billiard ball on the billiard table fails to track, the first movement track generated before the billiard ball fails to track is acquired, the second movement track generated after the billiard ball fails to track is determined based on the graph network, and the first movement track and the second movement track are correlated to be used as the complete movement track of the billiard ball, so that the problems that the image collected by a camera is blurred, the billiard ball is blocked by a user body, and the billiard ball is blocked by a ball rod, the multi-section movement track of the billiard ball cannot be correlated quickly, and the integrity of tracking the billiard ball is greatly improved.
The embodiment of the invention provides a multi-target tracking system based on a graph network, which comprises the following components: the device comprises a feature extraction module and a feature matching module.
The working principle of the technical scheme is as follows: the graph network has the functions of feature extraction and feature matching, and is not described in detail in the prior art.
The beneficial effects of the technical scheme are as follows: the graph network of the embodiment of the invention comprises a feature extraction module and a feature matching module.
The embodiment of the invention provides a multi-target tracking system based on a graph network, and the determining and associating module 4 performs the following operations:
acquiring a first image corresponding to a first time preset before tracking failure and a second image corresponding to a second time preset after tracking failure;
extracting a first feature of a tracking target on a first image through a feature extraction module;
extracting second features of each tracking target on the second image through a feature extraction module, and combining the second features into a second feature set;
selecting any one of the second features in the second feature set as a target feature;
and judging whether the first feature is matched with the target feature or not through a feature matching module, and if so, acquiring a second motion track generated by a tracking target corresponding to the target feature in a second time period.
The working principle of the technical scheme is as follows:
the preset first time is specifically: tracking the last moment before failure; the preset second moment is specifically: tracking the corresponding moments when the number of billiards (including clear recognition and fuzzy recognition) on the billiards table after failure meets a reasonable numerical value; for example: after a billiard ball is opened for a period of time, 10 billiards are left on the tabletop, after a certain billiard ball fails to track, no billiard ball falls into a bag during the period, and when 10 billiards appear on the tabletop again, the billiards are automatically tracked; the first and second features are specifically: appearance characteristics, spatial relationship characteristics and time relationship characteristics of billiards on the corresponding images.
The beneficial effects of the technical scheme are as follows: according to the embodiment of the invention, the second motion trail generated again by the tracking target (namely the billiard ball) in the preset second time period after the tracking failure is determined through the graph network, and the effect is accurate.
The embodiment of the invention provides a multi-target tracking system based on a graph network, which further comprises:
the judging and executing module is used for obtaining state parameters of the graph network, processing the state parameters to obtain graph network state scores, obtaining tracking parameters of each tracking target, processing the tracking parameters to obtain a table state index, judging whether the graph network state scores are matched with the table state index, and executing corresponding preset strategies if the graph network state scores are not matched with the table state index;
Wherein the status parameters include: the state value and the state prediction value of each node in the graph network, and the tracking parameters comprise: real-time position and real-time speed of each tracking target;
the judging and executing module executes the following operations:
calculating a graph network state score based on the state values and the state predictors:
wherein f is a table state index, v t For the t-th real-time speed of the tracking target, v 0 For a preset speed threshold, s t For the t-th tracking target real-time position, s e For the e-th real-time position of the tracking target, L is a preset distance function for obtaining the distance between the two positions, L 0 Z is the total number of the tracking targets and is a preset distance threshold;
calculating a table state index based on the real-time position and the real-time speed:
wherein f is a table state index, v t For the real-time speed of the t-th tracking target, v 0 For a preset speed threshold, s t For the t tracking target real-time position s e For the e-th tracking target real-time position, L is a preset distance function for obtaining the distance between the two positions, L 0 Is a preset distance threshold;
the judging and executing module judges whether the network state score of the graph is matched with the state index of the table by adopting the following preset method, and specifically comprises the following steps:
Acquiring a first interval corresponding to the network state score of the graph in a preset database of the state indexes of the suitable table;
acquiring a second interval corresponding to the table state index in a preset network state scoring interval database of the suitable graph;
calculating a matching index based on the first interval and the second interval:
wherein match is a matching index, f is a table state index, and alpha is a graph networkCollateral score, u 1 Is the upper limit value of the first interval, d 1 As the lower limit value of the first interval, u 2 Is the upper limit value of the second interval, d 2 Is the lower limit value of the second interval,and->The weight value is preset;
and when the matching index is smaller than or equal to a preset matching index threshold, determining that the network state score of the graph is not matched with the table state index.
The working principle of the technical scheme is as follows:
calculating a graph network state score based on a state value (such as a load flow value) and a state prediction value (such as a value for predicting a current state value after a large amount of historical operation data is learned through a pre-trained learning model) of each node in the graph network, wherein the lower the score is, the higher the load of the graph network is, and the worse the state is; the speed of billiards on a billiard table is high, when a plurality of billiards are close to the billiards, the probability of collision of the billiards is high, the billiards need to be ready for tracking, the situation of tracking loss is high due to the fact that the speed of each billiard is high at the moment of collision, and the total load of a graph network is high, therefore, the state index of the billiard table is calculated based on the real-time position and the real-time speed of each billiard, and the higher the state index is, the higher the probability of collision of the billiards is; when the state score of the graph network is not matched with the state index of the table, the graph network is insufficient in tracking loss processing performance, and a preset strategy (for example, a standby node is started) is required to be executed; the preset database suitable for the table state index interval is specifically: the table state index interval corresponding to each pre-stored network state score of each graph is, for example: if the network state score of the graph is 85, the corresponding appropriate table state index interval is [80,90]; the preset network state scoring interval database of the suitable graph is specifically: a network state scoring interval of a suitable map corresponding to each table state index is pre-stored, for example: if the table state index is 65, the corresponding network state scoring interval of the suitable graph is [60,70]; the preset matching index threshold value is specifically: for example 30.
The beneficial effects of the technical scheme are as follows: according to the embodiment of the invention, the running state of the graph network and the movement state of a plurality of billiards on the billiards table are respectively evaluated by calculating the graph network scores and the billiards table state indexes, when the graph network scores and the billiards table state indexes are not matched, the defect of insufficient processing performance of the graph network on tracking loss is illustrated, a preset strategy is required to be executed immediately, the fact that the speed of each billiard is higher at the moment of collision when a plurality of billiards are collided is ensured, and the processing performance of the graph network is greatly improved when the tracking loss is increased; and calculating a matching index based on the first interval and the second interval, and judging that the current network state score is not matched with the table state index when the matching index is smaller than or equal to a preset matching index threshold value, wherein the judgment is accurate.
The embodiment of the invention provides a multi-target tracking system based on a graph network, which further comprises:
the simulation module is used for simulating the simulated motion trail of any tracking target in time when the tracking failure time is greater than or equal to a preset first time threshold value;
wherein the simulation module performs operations comprising:
Taking the tracking target as a pre-simulation target;
acquiring the instantaneous position, the instantaneous speed and the instantaneous direction of a pre-simulated target at a third moment preset before tracking failure;
simulating a simulated motion trail of a pre-simulated target based on the instantaneous position, the instantaneous speed and the instantaneous direction;
the system further comprises:
the projection reminding module is used for projecting a corresponding preset reminding picture on the tabletop of the billiard table through the projection device to remind a user if the tracking failure time is greater than or equal to a preset second time threshold;
the projection reminding module performs the following operations:
acquiring a preset initial graph;
projecting the initial graph onto a desktop;
during the projection process, the side length of the initial graph is adjusted based on the time of tracking failure:
wherein L is the side length of the initial graph after determination, L 0 For the preset initial value of the side length, T is the time of tracking failure, T 0 Is a second time threshold.
The working principle of the technical scheme is as follows:
if the time of the failure tracking is greater than or equal to the preset first time threshold (for example, 5 seconds), if the billiard ball still moves in the period of time, the movement track of the billiard ball in the period of time is longer (two tracks can be directly connected in a shorter period of time), in order to display the whole track, a simulated movement track can be simulated based on the instant position, the instant speed and the instant direction of the billiard ball at the preset third moment (last moment before the failure tracking), and the simulated movement track can be fused with the first movement track and the second movement track to be used as the whole movement track of the billiard ball; if the tracking failure time is greater than or equal to a preset second time threshold (for example, 15 seconds), the user is informed that the unintentional shielding time is longer (for example, other users rely on the billiard table for a long time and the head and the like of the other users shield billiard balls), and the billiard ball is required to be reminded; the projection state is specifically: a projector; the preset initial graph is specifically: the frame is yellow, and the character of 'please remove the club/body and thank you' is arranged in the frame; if the tracking failure time is longer, the side length of the regular polygon can be adjusted, namely, the pattern range is enlarged, and a user is reminded to timely remove the club and the like.
The beneficial effects of the technical scheme are as follows: according to the embodiment of the invention, when the tracking failure time is greater than or equal to the first time threshold, the simulated motion track of the corresponding billiard ball in the period of the tracking failure is simulated, so that the track tracking integrity is improved.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims (8)

1. A graph network-based multi-target tracking method, comprising:
determining a plurality of tracking targets according to the operation instruction of the user;
tracking each tracking target according to a preset tracking method;
in the tracking process, if any tracking target fails to track, acquiring a first motion track generated by the tracking target within a first time period preset before the tracking failure;
Acquiring a second motion trail which is generated again by the tracking target in a second time period preset after the tracking failure based on a graph network, and correlating the first motion trail with the second motion trail to be used as a complete motion trail;
acquiring state parameters of the graph network, processing the state parameters to acquire graph network state scores, acquiring tracking parameters of each tracking target, processing the tracking parameters to acquire a table state index, judging whether the graph network state scores are matched with the table state index, and executing corresponding preset strategies if the graph network state scores are not matched with the table state index;
wherein the status parameters include: the state value and the state prediction value of each node in the graph network, and the tracking parameters comprise: real-time position and real-time speed of each tracking target;
the processing the state parameters to obtain a map network state score specifically includes:
calculating the graph network state score based on the state value and the state prediction value:
wherein α is the state score of the graph network, n is the total number of nodes in the graph network, x i Y, which is the state value of the ith node in the graph network i X is the state predictive value of the ith node in the graph network i,0 A threshold value of a preset state value corresponding to an ith node in the graph network, and is sum, or is or;
the processing the tracking parameters to obtain the table state index specifically includes:
calculating the table state index based on the real-time position and the real-time speed:
wherein f is a table state index, v t For the t-th real-time speed of the tracking target, v 0 For a preset speed threshold, s t For the t-th tracking target real-time position, s e For the e-th real-time position of the tracking target, L is a preset distance function for obtaining the distance between the two positions, L 0 Z is the total number of the tracking targets and is a preset distance threshold;
the judging whether the network state score of the graph is matched with the state index of the table specifically comprises the following steps:
acquiring a first interval corresponding to the network state score of the graph in a preset database of the state indexes of the suitable table;
acquiring a second interval corresponding to the table state index in a preset network state scoring interval database of the suitable graph;
calculating a matching index based on the first interval and the second interval:
Wherein match is a matching index, f is the table state index, α is the graph network score, u 1 D is the upper limit value of the first interval 1 U is the lower limit value of the first interval 2 D is the upper limit value of the second interval 2 As the lower limit value of the second interval,and->The weight value is preset;
and when the matching index is smaller than or equal to a preset matching index threshold, determining that the network state score of the graph and the table state index are not matched.
2. The multi-objective tracking method based on graph network of claim 1, wherein the graph network comprises: the device comprises a feature extraction module and a feature matching module.
3. The method for tracking multiple targets based on graph network according to claim 2, wherein the method for tracking multiple targets based on graph network comprises the steps of:
acquiring a first image corresponding to a first time preset before tracking failure and a second image corresponding to a second time preset after tracking failure;
extracting a first feature of the tracking target on the first image through the feature extraction module;
Extracting second features of each tracking target on the second image through the feature extraction module, and combining the second features into a second feature set;
selecting any one of the second features in the second feature set as a target feature;
judging whether the first feature is matched with the target feature or not through the feature matching module, and if so, acquiring the second motion trail generated by the tracking target corresponding to the target feature in the second time period.
4. The graph network-based multi-target tracking method of claim 1, further comprising:
when any tracking target fails to track, if the time of the tracking failure is greater than or equal to a preset first time threshold value, simulating the simulated motion trail of the tracking target in the time;
the simulating the simulated motion trail of the tracking target in the time specifically comprises the following steps:
taking the tracking target as a pre-simulation target;
acquiring the instantaneous position, the instantaneous speed and the instantaneous direction of the pre-simulation target at a third time preset before the tracking failure;
simulating a simulated motion trail of the pre-simulated target based on the instantaneous position, the instantaneous speed and the instantaneous direction;
The method further comprises the steps of:
if the tracking failure time is greater than or equal to a preset second time threshold, a corresponding preset reminding picture is projected on the tabletop of the billiard table through a projection device, and reminding is carried out on the user;
the method comprises the steps of projecting corresponding preset reminding pictures on a tabletop of a billiard table through a projection device, and specifically comprises the following steps:
acquiring a preset initial graph;
projecting the initial graphic onto the desktop;
during projection, adjusting the side length of the initial graph based on the time of the tracking failure:
wherein L is the determined side length of the initial graph, L 0 For the preset initial value of the side length, T is the time of failure of tracking, T 0 Is the second time threshold.
5. A graph network-based multi-target tracking system, comprising:
the determining module is used for determining a plurality of tracking targets according to the operation instruction of the user;
the tracking module is used for tracking each tracking target according to a preset tracking method;
the acquisition module is used for acquiring a first motion track generated by any tracking target in a first time period preset before the tracking failure if the tracking of the tracking target fails in the tracking process;
The determining and associating module is used for acquiring a second motion trail which is generated again by the tracking target in a second time period preset after the tracking failure based on the graph network, and associating the first motion trail with the second motion trail to be used as a complete motion trail;
the judging and executing module is used for obtaining state parameters of the graph network, processing the state parameters to obtain graph network state scores, obtaining tracking parameters of each tracking target, processing the tracking parameters to obtain a table state index, judging whether the graph network state scores are matched with the table state index, and executing corresponding preset strategies if the graph network state scores are not matched with the table state index;
wherein the status parameters include: the state value and the state prediction value of each node in the graph network, and the tracking parameters comprise: real-time position and real-time speed of each tracking target;
the judging and executing module executes the following operations:
calculating the graph network state score based on the state value and the state prediction value:
wherein α is the state score of the graph network, n is the total number of nodes in the graph network, x i Y, which is the state value of the ith node in the graph network i X is the state predictive value of the ith node in the graph network i,0 A preset state value threshold corresponding to an ith node in the graph network;
calculating the table state index based on the real-time position and the real-time speed:
wherein f is a table state index, v t For the t-th real-time speed of the tracking target, v 0 For a preset speed threshold, s t For the t-th tracking target real-time position, s e For the e-th real-time position of the tracking target, L is a preset distance function for obtaining the distance between the two positions, L 0 Z is the total number of the tracking targets and is a preset distance threshold;
the judging and executing module judges whether the network state score of the graph is matched with the state index of the table by adopting the following preset method, and specifically comprises the following steps:
acquiring a first interval corresponding to the network state score of the graph in a preset database of the state indexes of the suitable table;
acquiring a second interval corresponding to the table state index in a preset network state scoring interval database of the suitable graph;
calculating a matching index based on the first interval and the second interval:
wherein match is a matching index, f is the table state index, α is the graph network score, u 1 D is the upper limit value of the first interval 1 U is the lower limit value of the first interval 2 D is the upper limit value of the second interval 2 As the lower limit value of the second interval,and->The weight value is preset;
and when the matching index is smaller than or equal to a preset matching index threshold, determining that the network state score of the graph and the table state index are not matched.
6. The graph network-based multi-target tracking system of claim 5, wherein the graph network comprises: the device comprises a feature extraction module and a feature matching module.
7. The graph network-based multi-target tracking system of claim 6, wherein the determining and associating module performs operations comprising:
acquiring a first image corresponding to a first time preset before tracking failure and a second image corresponding to a second time preset after tracking failure;
extracting a first feature of the tracking target on the first image through the feature extraction module;
extracting second features of each tracking target on the second image through the feature extraction module, and combining the second features into a second feature set;
selecting any one of the second features in the second feature set as a target feature;
Judging whether the first feature is matched with the target feature or not through the feature matching module, and if so, acquiring the second motion trail generated by the tracking target corresponding to the target feature in the second time period.
8. The graph-network-based multi-target tracking system of claim 5, further comprising:
the simulation module is used for simulating the simulated motion trail of the tracking target in the time when the tracking failure time of any tracking target is greater than or equal to a preset first time threshold value;
wherein the simulation module performs operations comprising:
taking the tracking target as a pre-simulation target;
acquiring the instantaneous position, the instantaneous speed and the instantaneous direction of the pre-simulation target at a third time preset before the tracking failure;
simulating a simulated motion trail of the pre-simulated target based on the instantaneous position, the instantaneous speed and the instantaneous direction;
the system further comprises:
the projection reminding module is used for projecting a corresponding preset reminding picture on the tabletop of the billiard table through the projection device to remind a user if the tracking failure time is greater than or equal to a preset second time threshold;
The projection reminding module performs the following operations:
acquiring a preset initial graph;
projecting the initial graphic onto the desktop;
during projection, adjusting the side length of the initial graph based on the time of the tracking failure:
wherein L is the determined side length of the initial graph, L 0 For the preset initial value of the side length, T is the time of failure of tracking, T 0 Is the second time threshold.
CN202110104067.XA 2021-01-26 2021-01-26 Multi-target tracking method and system based on graph network Active CN112802067B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110104067.XA CN112802067B (en) 2021-01-26 2021-01-26 Multi-target tracking method and system based on graph network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110104067.XA CN112802067B (en) 2021-01-26 2021-01-26 Multi-target tracking method and system based on graph network

Publications (2)

Publication Number Publication Date
CN112802067A CN112802067A (en) 2021-05-14
CN112802067B true CN112802067B (en) 2024-01-26

Family

ID=75811833

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110104067.XA Active CN112802067B (en) 2021-01-26 2021-01-26 Multi-target tracking method and system based on graph network

Country Status (1)

Country Link
CN (1) CN112802067B (en)

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106780620A (en) * 2016-11-28 2017-05-31 长安大学 A kind of table tennis track identification positioning and tracking system and method
CN109448025A (en) * 2018-11-09 2019-03-08 国家体育总局体育科学研究所 Short-track speeding skating sportsman's automatically tracks and track modeling method in video
CN109584265A (en) * 2017-09-28 2019-04-05 杭州海康威视数字技术股份有限公司 A kind of method for tracking target and device
CN109903312A (en) * 2019-01-25 2019-06-18 北京工业大学 A kind of football sportsman based on video multi-target tracking runs distance statistics method
CN110660084A (en) * 2019-09-30 2020-01-07 上海淡竹体育科技有限公司 Multi-target tracking method and device
CN110660083A (en) * 2019-09-27 2020-01-07 国网江苏省电力工程咨询有限公司 Multi-target tracking method combined with video scene feature perception
CN110807377A (en) * 2019-10-17 2020-02-18 浙江大华技术股份有限公司 Target tracking and intrusion detection method, device and storage medium
CN111429483A (en) * 2020-03-31 2020-07-17 杭州博雅鸿图视频技术有限公司 High-speed cross-camera multi-target tracking method, system, device and storage medium
CN111639551A (en) * 2020-05-12 2020-09-08 华中科技大学 Online multi-target tracking method and system based on twin network and long-short term clues
CN111696128A (en) * 2020-05-27 2020-09-22 南京博雅集智智能技术有限公司 High-speed multi-target detection tracking and target image optimization method and storage medium
CN111914664A (en) * 2020-07-06 2020-11-10 同济大学 Vehicle multi-target detection and track tracking method based on re-identification

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101231755B (en) * 2007-01-25 2013-03-06 上海遥薇(集团)有限公司 Moving target tracking and quantity statistics method
CN105046059B (en) * 2015-06-24 2017-09-29 深圳市腾讯计算机系统有限公司 Play sport track determines method and device

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106780620A (en) * 2016-11-28 2017-05-31 长安大学 A kind of table tennis track identification positioning and tracking system and method
CN109584265A (en) * 2017-09-28 2019-04-05 杭州海康威视数字技术股份有限公司 A kind of method for tracking target and device
CN109448025A (en) * 2018-11-09 2019-03-08 国家体育总局体育科学研究所 Short-track speeding skating sportsman's automatically tracks and track modeling method in video
CN109903312A (en) * 2019-01-25 2019-06-18 北京工业大学 A kind of football sportsman based on video multi-target tracking runs distance statistics method
CN110660083A (en) * 2019-09-27 2020-01-07 国网江苏省电力工程咨询有限公司 Multi-target tracking method combined with video scene feature perception
CN110660084A (en) * 2019-09-30 2020-01-07 上海淡竹体育科技有限公司 Multi-target tracking method and device
CN110807377A (en) * 2019-10-17 2020-02-18 浙江大华技术股份有限公司 Target tracking and intrusion detection method, device and storage medium
CN111429483A (en) * 2020-03-31 2020-07-17 杭州博雅鸿图视频技术有限公司 High-speed cross-camera multi-target tracking method, system, device and storage medium
CN111639551A (en) * 2020-05-12 2020-09-08 华中科技大学 Online multi-target tracking method and system based on twin network and long-short term clues
CN111696128A (en) * 2020-05-27 2020-09-22 南京博雅集智智能技术有限公司 High-speed multi-target detection tracking and target image optimization method and storage medium
CN111914664A (en) * 2020-07-06 2020-11-10 同济大学 Vehicle multi-target detection and track tracking method based on re-identification

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
"Online multi-target tracking using non-linear motion estimation";Sijong Kim;《 2014 11th International Conference on Ubiquitous Robots and Ambient Intelligence》;全文 *
车辆运动轨迹建模和跟踪滤波;季思文;闫胜业;黄宇维;;计算机仿真(第10期);全文 *

Also Published As

Publication number Publication date
CN112802067A (en) 2021-05-14

Similar Documents

Publication Publication Date Title
CN108256433B (en) Motion attitude assessment method and system
CN105069472B (en) A kind of vehicle checking method adaptive based on convolutional neural networks
Datta et al. Person-on-person violence detection in video data
CN109343700B (en) Eye movement control calibration data acquisition method and device
US9183431B2 (en) Apparatus and method for providing activity recognition based application service
CN110765964A (en) Method for detecting abnormal behaviors in elevator car based on computer vision
CN110287907B (en) Object detection method and device
CN106846362A (en) A kind of target detection tracking method and device
CN114100103B (en) Rope skipping counting detection system and method based on key point identification
CN109544592A (en) For the mobile moving object detection algorithm of camera
CN111860091A (en) Face image evaluation method and system, server and computer readable storage medium
CN101826155B (en) Method for identifying act of shooting based on Haar characteristic and dynamic time sequence matching
CN108960098A (en) A kind of billiard ball collision relation recognition method and game of billiards scoring system
CN113888588A (en) Target tracking method, device, equipment and storage medium
CN109636846A (en) Object localization method based on circulation attention convolutional neural networks
Spagnolo et al. Non-invasive soccer goal line technology: a real case study
CN112802066B (en) Multi-target tracking method and system based on multi-track fusion
Sokolova et al. Human identification by gait from event-based camera
CN112802067B (en) Multi-target tracking method and system based on graph network
Ju et al. Trajectory-based Badminton Shots Detection
CN107730532B (en) Badminton motion trajectory tracking method, system, medium and equipment
CN115713714A (en) Motion evaluation method, device, system, electronic device and storage medium
CN106139542B (en) Golf shot trigger and its method for sensing
CN113836980A (en) Face recognition method, electronic device and storage medium
CN112598707A (en) Real-time video stream object detection and tracking method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant