A wearable athlete posture analysis method and system

Abstract

The present invention discloses a wearable athlete posture analysis method and system, the method includes: posture data collection, construction of abnormal posture recognition model, optimization of sports performance, prevention of injury and personalized training. The present invention belongs to the field of intelligent sports technology, specifically refers to a wearable athlete posture analysis method and system, the scheme adopts the construction of abnormal posture recognition model, converts high-dimensional data information into low-dimensional, calculates the target function of segmentation, constructs the initial tree core, and optimizes the generation of the tree core, accurately identifies the abnormal posture of athletes, and enhances the training effect; the training arrangement is adjusted to adapt to the individual differences of athletes, the training data representation is obtained by using random exchange and random missing technology, the posture intention is captured by a multi-head self-attention network, the posture data is selected according to the training direction, the personalized training results are output, the appropriate training method is selected, and the personal potential is explored.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent sports technology, and specifically relates to a wearable athlete posture analysis method and system. Background Art

[0002] The athlete posture analysis method is a method that uses wearable devices to collect athletes' sports data and evaluates and analyzes athletes' postures through data analysis and processing. It aims to monitor athletes' body postures, movement ranges and movement trajectories in real time, and help athletes and coaches better understand athletes' performance and potential problems so as to conduct targeted training and improvement. However, the existing athlete posture analysis methods have technical problems such as inaccurate recognition of athletes' abnormal postures, failure to timely discover incorrect training movements, affecting training quality and improvement effects, and continuing to exercise in wrong postures, increasing the chance of injury; and technical problems such as the diversity of athletes' situations, if personalized training cannot be set up, it will be difficult for athletes to make substantial progress in a training mode that is not suitable for them, limiting the potential of athletes. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a wearable athlete posture analysis method and system, which aims at the technical problem that the abnormal posture recognition of athletes is inaccurate, resulting in the failure to timely discover incorrect training movements, affecting the training quality and improvement effect, continuing to exercise in wrong postures, and increasing the chance of injury. An abnormal posture recognition model is constructed to convert high-dimensional data information into low-dimensional, calculate the objective function of segmentation, construct an initial tree core, and optimize the generation of the tree core; in view of the fact that there are diverse athlete situations, if personalized training cannot be set, it will lead to athletes. It is difficult to make substantial progress in a training mode that is not suitable for them, which limits the potential of athletes. The training arrangement is adjusted to adapt to the individual differences of athletes, and the training data representation is obtained by using random exchange and random missing technology. The posture intention is captured through a multi-head self-attention network, the posture data is selected according to the training direction, and the personalized training results are output.

[0004] The technical solution adopted by the present invention is as follows: The present invention provides a wearable athlete posture analysis method, which comprises the following steps:

[0005] Step S1: posture data collection, specifically collecting the athlete's posture data;

[0006] Step S2: construct an abnormal posture recognition model, specifically, perform feature extraction, convert high-dimensional data information into low-dimensional, calculate the target function of segmentation, construct an initial tree core, and optimize the generation of the tree core;

[0007] Step S3: Optimizing sports performance, specifically providing professional action guidance based on abnormal posture recognition results;

[0008] Step S4: Preventing injuries, specifically continuous monitoring, setting up early warning mechanisms, and wearing protective equipment;

[0009] Step S5: Personalized training, specifically adjusting the training arrangement to adapt to the individual differences of athletes, using random exchange and random missing techniques to obtain training data representation, capturing posture intentions through a multi-head self-attention network, selecting posture data according to the training direction, and outputting personalized training results.

[0010] Furthermore, in step S1, the posture data is collected by placing a pressure sensor on the sole of the shoe to collect pressure distribution data of the athlete under different movements, and attaching an electromyography sensor to the muscle surface to obtain the electrical signal generated during muscle activity to reflect the muscle force.

[0011] Furthermore, in step S2, the abnormal posture recognition model is constructed, including the following steps:

[0012] Step S21: feature extraction, performing feature extraction to convert posture data into feature vectors for model training;

[0013] Step S22: Convert high-dimensional data information into low-dimensional data, greatly reducing the complexity of data processing and improving the accuracy and timeliness of gesture recognition;

[0014] Step S23: Calculate the objective function of segmentation, divide the posture data after dimension reduction into different categories, and better identify abnormal postures. The formula used is as follows:

[0015] ;

[0016] In the formula, M represents the objective function, U represents the feature vector set of the cluster center, represents the dynamic evolution pattern vector set contained in the feature, i is the index of the data cluster, L is the total number of data clusters, j is the index of the posture data point, e is the total number of posture data points, The regularization term representing the number of partitions is used to prevent overfitting in dynamic evolution. represents the L2 norm, λ is the balancing factor, represents the smoothed eigenvector, represents the pose sample data point, The elements in the eigenvector set representing the cluster center, Transpose the elements in the eigenvector set representing the cluster center;

[0017] Step S24: constructing an initial tree core, after completing the posture data classification according to the objective function, constructing different nodes and hierarchical structures in the initial tree core according to the classification results, and reflecting the athlete's posture information according to the joint position, speed and angle change feature vectors contained in the category;

[0018] Step S25: Optimize the tree core generation to further strengthen the correlation between actions, improve the abnormal posture recognition effect, and calculate the optimized equation. The formula used is as follows:

[0019] ;

[0020] In the formula, Represents the elements of the dynamically evolving pattern vector set contained in the feature.

[0021] Furthermore, in step S3, the optimization of sports performance is specifically to train and enhance the core muscle strength according to the abnormal posture recognition results so as to better stabilize and control the posture. According to the information provided by the model, the coach gives the athlete professional technical guidance, including limb position and movement sequence.

[0022] Further, in step S4, the damage prevention comprises the following steps:

[0023] Step S41: Continuous monitoring, by monitoring the abnormal posture recognition model, timely discovering abnormal posture tendencies that may cause injury;

[0024] Step S42: Setting up an early warning mechanism to immediately send out an alarm when an athlete shows abnormal posture tendency, provide feedback on training content and intensity arrangement, avoid excessive training leading to increased risk of injury, and provide guidance for rehabilitation training for athletes with a history of injury;

[0025] Step S43: Wear protective equipment, understand the athlete's posture characteristics according to the model, and select wearable equipment to provide better protection and support.

[0026] Furthermore, in step S5, the personalized training includes the following steps:

[0027] Step S51: Obtain training data, use random exchange and random missing techniques to increase the diversity and complexity of data, improve generalization ability, and obtain training data representation. The formula used is as follows:

[0028] ;

[0029] In the formula, S represents the training data after random exchange and random missing technology, that is, enhanced training data, is the first enhancement weight, is the second enhancement weight, RS represents the random swap operation, RD represents the random deletion operation, Represents the original training data;

[0030] Step S52: Capture posture intention, apply a multi-head self-attention network, inject the original training data and enhanced training data into different attention heads at the same time, to help capture the intention behind the athlete's posture and potential action guidance. The formula used is as follows:

[0031] ;

[0032] ;

[0033] In the formula, represents the attention weight, Represents a specific posture data point, represents the parameters of the multi-head self-attention network, represents the feature representation after calculation, represents the characteristic representation factor;

[0034] Step S53: Select posture data according to the training direction, and use the additive attention mechanism to select posture data that better represents the athlete's training direction. The formula used is as follows:

[0035] ;

[0036] ;

[0037] In the formula, Indicates the training direction result, represents the training direction parameter vector, tanh represents the hyperbolic tangent function, represents the additive attention mechanism parameters, represents the additive factor, Indicates the importance of posture data in the training direction, exp represents the exponential function, Represents the training result of the j-th posture data point;

[0038] Step S54: output the personalized training result, the formula used is as follows:

[0039] ;

[0040] In the formula, u represents the representation of personalized training results;

[0041] Step S55: Adjust the training arrangement according to the personalized training results, analyze the personalized training results, including the specific data of each indicator, progress and existing problems, and re-plan the training content, such as adjusting the order of training items, increasing the training frequency of specific items and replacing training equipment, so as to better adapt to the individual differences of athletes and improve the training effect. Dynamically adjust the training arrangement according to the specific situation of the athletes, avoid ineffective training, and help athletes improve their abilities and performance more quickly.

[0042] The present invention provides a wearable athlete posture analysis system, which includes a posture data acquisition module, an abnormal posture recognition model building module, a sports performance optimization module, an injury prevention module and a personalized training module;

[0043] The posture data collection module is specifically used to collect the posture data of athletes;

[0044] The abnormal posture recognition model module is constructed, specifically to extract features, convert high-dimensional data information into low-dimensional, calculate the target function of segmentation, construct an initial tree core, and optimize the generation of the tree core;

[0045] The sports performance optimization module specifically provides professional action guidance based on the abnormal posture recognition results;

[0046] The injury prevention module specifically includes continuous monitoring, setting up early warning mechanisms, and wearing protective equipment;

[0047] The personalized training module specifically adjusts the training arrangement to adapt to the individual differences of athletes, uses random exchange and random deletion technology to obtain training data representation, captures posture intentions through a multi-head self-attention network, selects posture data according to the training direction, and outputs personalized training results.

[0048] The beneficial results achieved by the present invention using the above scheme are as follows:

[0049] (1) In order to solve the technical problem that athletes’ abnormal posture recognition is inaccurate, which leads to the inability to timely discover incorrect training movements, affects the training quality and improvement effect, and continues to exercise in the wrong posture, increasing the chance of injury, we build an abnormal posture recognition model, convert high-dimensional data information into low-dimensional, calculate the objective function of segmentation, build the initial tree core, and optimize the generation of the tree core;

[0050] (2) In view of the diverse situations of athletes, if personalized training cannot be set up, it will be difficult for athletes to make substantial progress under a training mode that is not suitable for them, which will limit the technical problems of athletes' potential. The training schedule is adjusted to adapt to the individual differences of athletes, and the training data representation is obtained using random exchange and random missing techniques. The posture intention is captured through a multi-head self-attention network, and the posture data is selected according to the training direction to output personalized training results. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A schematic flow chart of a wearable athlete posture analysis method provided by the present invention;

[0052] Figure 2 A schematic diagram of a wearable athlete posture analysis system provided by the present invention;

[0053] Figure 3 is a schematic flow chart of step S2;

[0054] Figure 4 It is a flowchart diagram of step S5.

[0055] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0056] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0057] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0058] Example 1, see Figure 1 The present invention provides a wearable athlete posture analysis method, which comprises the following steps:

[0059] Step S1: posture data collection, specifically collecting the athlete's posture data;

[0060] Step S2: construct an abnormal posture recognition model, specifically, perform feature extraction, convert high-dimensional data information into low-dimensional, calculate the target function of segmentation, construct an initial tree core, and optimize the generation of the tree core;

[0061] Step S3: Optimizing sports performance, specifically providing professional action guidance based on abnormal posture recognition results;

[0062] Step S4: Preventing injuries, specifically continuous monitoring, setting up early warning mechanisms, and wearing protective equipment;

[0063] Step S5: Personalized training, specifically adjusting the training arrangement to adapt to the individual differences of athletes, using random exchange and random missing techniques to obtain training data representation, capturing posture intentions through a multi-head self-attention network, selecting posture data according to the training direction, and outputting personalized training results.

[0064] Example 2, see Figure 1 This embodiment is based on the above embodiment. In step S1, the posture data is collected by placing a pressure sensor on the sole of the shoe to collect pressure distribution data of the athlete under different actions, and attaching an electromyography sensor to the muscle surface to obtain the electrical signal generated during muscle activity to reflect the muscle force.

[0065] Example 3, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. In step S2, the abnormal posture recognition model is constructed, including the following steps:

[0066] Step S21: feature extraction, performing feature extraction to convert posture data into feature vectors for model training;

[0067] Step S22: Convert high-dimensional data information into low-dimensional data, greatly reducing the complexity of data processing and improving the accuracy and timeliness of gesture recognition;

[0068] Step S23: Calculate the objective function of segmentation, divide the posture data after dimension reduction into different categories, and better identify abnormal postures. The formula used is as follows:

[0069] ;

[0070] In the formula, M represents the objective function, U represents the feature vector set of the cluster center, represents the dynamic evolution pattern vector set contained in the feature, i is the index of the data cluster, L is the total number of data clusters, j is the index of the posture data point, e is the total number of posture data points, The regularization term representing the number of partitions is used to prevent overfitting in dynamic evolution. represents the L2 norm, λ is the balancing factor, represents the smoothed eigenvector, represents the pose sample data point, The elements in the eigenvector set representing the cluster center, Transpose the elements in the eigenvector set representing the cluster center;

[0071] Step S24: constructing an initial tree core, after completing the posture data classification according to the objective function, constructing different nodes and hierarchical structures in the initial tree core according to the classification results, and reflecting the athlete's posture information according to the joint position, speed and angle change feature vectors contained in the category;

[0072] Step S25: Optimize the tree core generation to further strengthen the correlation between actions, improve the abnormal posture recognition effect, and calculate the optimized equation. The formula used is as follows:

[0073] ;

[0074] In the formula, Represents the elements of the dynamically evolving pattern vector set contained in the feature.

[0075] By performing the above operations, an abnormal posture recognition model is constructed, high-dimensional data information is converted into low-dimensional, the objective function of segmentation is calculated, the initial tree core is constructed, and the generation of the tree core is optimized. The technical problem of inaccurate recognition of athletes' abnormal postures, failure to timely discover incorrect training movements, affecting training quality and improvement effects, and continuing to exercise in wrong postures, increasing the chance of injury is solved.

[0076] Example 4, see Figure 1 This embodiment is based on the above embodiment. In step S3, the optimization of sports performance is specifically to train and enhance the core muscle strength according to the abnormal posture recognition results so as to better stabilize and control the posture. According to the information provided by the model, the coach gives the athlete professional technical guidance, including limb position and movement sequence.

[0077] Example 5, see Figure 1 This embodiment is based on the above embodiment. In step S4, the damage prevention comprises the following steps:

[0078] Step S41: Continuous monitoring, by monitoring the abnormal posture recognition model, timely discovering abnormal posture tendencies that may cause injury;

[0079] Step S42: Setting up an early warning mechanism to immediately send out an alarm when an athlete shows abnormal posture tendency, provide feedback on training content and intensity arrangement, avoid excessive training leading to increased risk of injury, and provide guidance for rehabilitation training for athletes with a history of injury;

[0080] Step S43: Wear protective equipment, understand the athlete's posture characteristics according to the model, and select wearable equipment to provide better protection and support.

[0081] Example 6, see Figure 1 and Figure 4 This embodiment is based on the above embodiment. In step S5, the personalized training includes the following steps:

[0082] Step S51: Obtain training data, use random exchange and random missing techniques to increase the diversity and complexity of data, improve generalization ability, and obtain training data representation. The formula used is as follows:

[0083] ;

[0084] In the formula, S represents the training data after random exchange and random missing technology, that is, enhanced training data, is the first enhancement weight, is the second enhancement weight, RS represents the random swap operation, RD represents the random deletion operation, Represents the original training data;

[0085] Step S52: Capture posture intention, apply a multi-head self-attention network, inject the original training data and enhanced training data into different attention heads at the same time, to help capture the intention behind the athlete's posture and potential action guidance. The formula used is as follows:

[0086] ;

[0087] ;

[0088] In the formula, represents the attention weight, Represents a specific posture data point, represents the parameters of the multi-head self-attention network, represents the feature representation after calculation, represents the characteristic representation factor;

[0089] Step S53: Select posture data according to the training direction, and use the additive attention mechanism to select posture data that better represents the athlete's training direction. The formula used is as follows:

[0090] ;

[0091] ;

[0092] In the formula, Indicates the training direction result, represents the training direction parameter vector, tanh represents the hyperbolic tangent function, represents the additive attention mechanism parameters, represents the additive factor, Indicates the importance of posture data in the training direction, exp represents the exponential function, Represents the training result of the j-th posture data point;

[0093] Step S54: output the personalized training result, the formula used is as follows:

[0094] ;

[0095] In the formula, u represents the representation of personalized training results;

[0096] Step S55: Adjust the training arrangement according to the personalized training results, analyze the personalized training results, including the specific data of each indicator, progress and existing problems, and re-plan the training content, such as adjusting the order of training items, increasing the training frequency of specific items and replacing training equipment, so as to better adapt to the individual differences of athletes and improve the training effect. Dynamically adjust the training arrangement according to the specific situation of the athletes, avoid ineffective training, and help athletes improve their abilities and performance more quickly.

[0097] By performing the above operations, the training schedule is adjusted to adapt to the individual differences of athletes, the random exchange and random missing techniques are used to obtain training data representation, the posture intention is captured through a multi-headed self-attention network, the posture data is selected according to the training direction, and the personalized training results are output. This solves the problem that the athletes are diverse in situation. If personalized training cannot be set, it will lead to athletes finding it difficult to make substantial progress in a training mode that is not suitable for them, which will limit the technical problem of athletes' potential.

[0098] Embodiment 7, see Figure 2 , this embodiment is based on the above embodiment, and the present invention provides a wearable athlete posture analysis system, including a posture data acquisition module, an abnormal posture recognition model building module, a sports performance optimization module, an injury prevention module and a personalized training module;

[0099] The posture data collection module is specifically used to collect the posture data of athletes;

[0100] The abnormal posture recognition model module is constructed, specifically to extract features, convert high-dimensional data information into low-dimensional, calculate the target function of segmentation, construct an initial tree core, and optimize the generation of the tree core;

[0101] The sports performance optimization module specifically provides professional action guidance based on the abnormal posture recognition results;

[0102] The injury prevention module specifically includes continuous monitoring, setting up early warning mechanisms, and wearing protective equipment;

[0103] The personalized training module specifically adjusts the training arrangement to adapt to the individual differences of athletes, uses random exchange and random deletion technology to obtain training data representation, captures posture intentions through a multi-head self-attention network, selects posture data according to the training direction, and outputs personalized training results.

[0104] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0105] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0106] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A wearable athlete posture analysis method, characterized by: The method comprises the following steps: Step S1: posture data collection; Step S2: construct an abnormal posture recognition model, specifically, perform feature extraction, convert high-dimensional data information into low-dimensional, calculate the target function of segmentation, construct an initial tree core, and optimize the generation of the tree core; Step S3: Optimizing sports performance, specifically providing professional action guidance based on abnormal posture recognition results; Step S4: prevent injury, specifically, continuous monitoring, timely discover abnormal posture tendencies through monitoring of abnormal posture recognition models, set up early warning mechanisms, and wear protective equipment; Step S5: personalized training, specifically obtaining training data, using random exchange and random missing techniques to obtain training data representation, capturing posture intentions through a multi-head self-attention network, selecting posture data according to the training direction, calculating weighted sums and outputting personalized training results; In step S2, the abnormal posture recognition model is constructed, including the following steps: Step S21: feature extraction, performing feature extraction to convert posture data into feature vectors for model training; Step S22: converting high-dimensional data information into low-dimensional; Step S23: Calculate the target function of segmentation, the formula used is as follows: ； In the formula, M represents the objective function, U represents the feature vector set of the cluster center, represents the dynamic evolution pattern vector set contained in the feature, i is the index of the data cluster, L is the total number of data clusters, j is the index of the posture data point, e is the total number of posture data points, The regularization term representing the number of partitions is used to prevent overfitting in dynamic evolution. represents the L2 norm, λ is the balancing factor, represents the smoothed eigenvector, represents the pose sample data point, The elements in the eigenvector set representing the cluster center, Transpose the elements in the eigenvector set representing the cluster center; Step S24: constructing an initial tree core, after completing the posture data classification according to the objective function, constructing different nodes and hierarchical structures in the initial tree core according to the classification results, and reflecting the athlete's posture information according to the joint position, speed and angle change feature vectors contained in the category; Step S25: Optimize the tree core generation and calculate the optimized equation. The formula used is as follows: ； In the formula, Represents the elements of the dynamically evolving pattern vector set contained in the feature.

2. A wearable athlete posture analysis method according to claim 1, characterized in that: In step S5, the personalized training includes the following steps: Step S51: Obtain training data, and use random exchange and random deletion techniques to obtain training data representation. The formula used is as follows: ； In the formula, S represents the training data after random exchange and random missing technology processing, is the first enhancement weight, is the second enhancement weight, RS represents the random swap operation, RD represents the random deletion operation, represents the original training data; Step S52: Capture the posture intention, apply the multi-head self-attention network, and inject the original posture data and the enhanced posture data into different attention heads at the same time. The formula used is as follows: ； ； In the formula, represents the attention weight, Represents a specific posture data point, represents the parameters of the multi-head self-attention network, represents the feature representation after calculation, represents the characteristic representation factor; Step S53: Select posture data according to the training direction, and use the additive attention mechanism to select posture data that better represents the athlete's training direction. The formula used is as follows: ； ； In the formula, Indicates the training direction result, represents the training direction parameter vector, tanh represents the hyperbolic tangent function, represents the additive attention mechanism parameters, represents the additive factor, Indicates the importance of posture data in the training direction, exp represents the exponential function, Represents the training result of the j-th posture data point; Step S54: output the personalized training result, the formula used is as follows: ； In the formula, u represents the representation of personalized training results; Step S55: Adjust the training schedule.

3. A wearable athlete posture analysis method according to claim 1, characterized in that: In step S1, the posture data is collected by placing a pressure sensor on the sole of the shoe to collect pressure distribution data of the athlete under different movements, and attaching an electromyography sensor to the muscle surface to obtain the electrical signal generated during muscle activity to reflect the muscle force.

4. A wearable athlete posture analysis method according to claim 1, characterized in that: In step S3, the optimization of sports performance is specifically to train and enhance the core muscle strength according to the abnormal posture recognition result, and the coach gives professional technical guidance to the athlete.

5. A wearable athlete posture analysis method according to claim 1, characterized in that: In step S4, the damage prevention comprises the following steps: Step S41: Continuously monitor, and timely discover abnormal posture tendencies by monitoring the abnormal posture recognition model; Step S42: Setting up an early warning mechanism to immediately send out an alarm when an athlete shows abnormal posture tendency, provide feedback on training content and intensity arrangement, and provide guidance on rehabilitation training for athletes with a history of injuries; Step S43: Wear protective equipment, understand the athlete's posture characteristics according to the model, and select wearable equipment to provide better protection and support.

6. A wearable athlete posture analysis system, used to implement a wearable athlete posture analysis method as claimed in any one of claims 1 to 5, characterized in that: It includes posture data collection module, abnormal posture recognition model building module, sports performance optimization module, injury prevention module and personalized training module.

7. A wearable athlete posture analysis system according to claim 6, characterized in that: The posture data collection module is specifically used to collect the posture data of athletes; The abnormal posture recognition model module is constructed, specifically to extract features, convert high-dimensional data information into low-dimensional, calculate the target function of segmentation, construct an initial tree core, and optimize the generation of the tree core; The sports performance optimization module specifically provides professional action guidance according to the output results of the abnormal posture recognition model; The injury prevention module specifically includes continuous monitoring, setting up early warning mechanisms, and wearing protective equipment; The personalized training module specifically adjusts the training arrangement to adapt to the individual differences of athletes, uses random exchange and random deletion technology to obtain training data representation, captures posture intentions through a multi-head self-attention network, selects posture data according to the training direction, and outputs personalized training results.

Images