CN118015712B - Tai Chi scoring method, system, device and medium based on time window - Google Patents

Abstract

The invention discloses a Tai Chi scoring method based on a time window, a system device and a medium, comprising: collecting Tai Chi action key points in a Tai Chi video, and defining the key points in each frame number in the video; taking the actions and gestures in the defined key points as a set of key data, constructing a mathematical matrix, and obtaining a recognition model through the mathematical matrix; according to the recognition model, dividing the actions and gestures in the mathematical matrix using a time window, dynamically adjusting the actions and gestures after the time window division by setting an adjustment coefficient, and outputting a model result; the invention divides the actions and gestures in the mathematical matrix using a time window, determines key frames, and thus realizes accurate division of the actions; dynamically adjusting the actions and gestures after the time window division by setting an adjustment coefficient, and outputting a model result, thereby effectively improving the robustness of the model and ensuring the accuracy of the Tai Chi score output by the model.

Description

Tai Chi scoring method, system, device and medium based on time window

Technical Field

The present invention relates to the technical field of Tai Chi scoring, and in particular to a Tai Chi scoring method, system, device and medium based on a time window.

Background technique

Nowadays, Tai Chi is highly regarded as an elective course in physical education in colleges and universities because it contains profound Chinese cultural traditions and can affect a person's physical and mental health, making it a popular physical education course.

Some people have suggested that the video can be used to analyze and score Tai Chi movements frame by frame to evaluate the correctness of the movements. This method requires the use of models for training and analysis, while the deep learning model training in traditional methods requires a lot of computing resources and requires the definition of evaluation criteria and features in advance, which limits the flexibility in dealing with personalized and non-traditional Tai Chi performance styles.

Summary of the invention

The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

In view of the above existing problems, the present invention is proposed. Therefore, the present invention provides a Tai Chi scoring method based on a time window, which uses the time window to divide the actions and gestures in the mathematical matrix, determines the key frames or feature points, and thus realizes the accurate division of the actions; by setting the adjustment coefficient, the actions and gestures after the time window division are dynamically adjusted, and the model results are output, which effectively improves the robustness of the model and ensures the accuracy of the Tai Chi score output by the model.

In order to achieve the above object, the present invention adopts the following technical solutions:

In a first aspect, the present invention provides a Tai Chi scoring method based on a time window, comprising:

Collect key points of Tai Chi movements in Tai Chi videos, and define key points in each frame of the video, wherein the key points include coordinate information and gesture marks;

The coordinate information and gesture marks of the defined key points are used as a set of key data to construct a mathematical matrix, and a recognition model is obtained through the mathematical matrix; each column corresponds to the four-dimensional information of a key point;

According to the recognition model, the actions and gestures in the mathematical matrix are divided by using a time window, the actions and gestures after the time window division are dynamically adjusted by setting an adjustment coefficient, and a model result is output;

According to the output results of the recognition model, the model is evaluated as a whole to obtain the Tai Chi scoring results.

As a preferred technical solution, key points of Tai Chi movements in Tai Chi videos are collected, and each frame number in the key points is defined, including:

For each frame, there are several key points, each of which contains coordinate information and gesture markers;

There are N key points, and the information of each key point is represented as a four-dimensional vector (x, y, t, g);

Among them, x and y are the coordinates of the key point in the Tai Chi video, t is the timestamp corresponding to the Tai Chi video, and g is the gesture marker.

As a preferred technical solution, the actions and gestures in the defined key points are used as a set of key data to construct a mathematical matrix, and a recognition model is obtained through the mathematical matrix, including:

Set the number of frames to M and express the matrix as , construct the mathematical matrix X, expressed as:

;

Among them, X represents a multidimensional matrix composed of M frames and N key points. , , , Respectively represent the abscissa, ordinate, time and marked gesture of the first key point in the first frame; , , , Respectively represent the abscissa, ordinate, time and marked gesture of the second key point in the first frame; , , , Respectively represent the horizontal coordinate, vertical coordinate, time and marked gesture of the Nth key point in the first frame; , , , Respectively represent the abscissa, ordinate, time and marked gesture of the first key point in the second frame; , , , Respectively represent the horizontal coordinate, vertical coordinate, time and marked gesture of the Nth key point in the Mth frame;

Inputting the X into the recognition model f, we get:

;

in, and They are the recognition results of actions and gestures respectively.

As a preferred technical solution, when the number of frames is input into the mathematical matrix X, it is necessary to delete the continuous frames whose key point detection number does not reach the set threshold and the image frames whose key point detection number does not meet the set threshold, and adjust the timestamp t corresponding to the Tai Chi video.

As a preferred technical solution, according to the recognition model, the actions and gestures in the mathematical matrix are divided using a time window, including:

The time window sequence in the recognition model is: , the gesture sequence is ,

Among them, each represents the i-th time window in the video, each represents the mth gesture;

The start frame of the gesture sequence The starting frame corresponding to the time window sequence , starting from the frame after the start frame of the time window, the current frame of each frame of the gesture sequence is the start frame of the time window of the next frame, until the current frame of the gesture sequence is equal to the start frame of the time window, and the time window sequence is obtained .

As a preferred technical solution, by setting the adjustment coefficient, the actions and gestures after the time window division are dynamically adjusted, and the model results are output, including:

Setting dual adjustment coefficients and ;

pass For time window series Adjust to obtain the change sequence corresponding to the gesture sequence ; In the change sequence On this basis, the recognition model is adjusted to obtain:

;

in, Indicates that at time point t Adjusted time series values; is the time window series adjustment coefficient, For the change sequence Dynamic coefficient of represents the time period for identifying the model, The score of Tai Chi output by the recognition model.

As a preferred technical solution, it also includes:

Define the model loss function and introduce binary variables The output result of the recognition model is judged; the loss function includes an action loss function and a gesture loss function;

If the binary variable more than the , it is marked as 1, which is a trustworthy model; otherwise, it is marked as 0, which is an untrustworthy model, and returns to the construction of the mathematical matrix to rebuild the model.

In a second aspect, the present invention provides a time window-based Tai Chi scoring system, based on the time window-based Tai Chi scoring method, comprising:

A video acquisition module, which acquires key points of Tai Chi movements in Tai Chi videos and defines key points in each frame of the video, wherein the key points include coordinate information and gesture marks;

The model building module uses the coordinate information and gesture marks of the defined key points as a set of key data to construct a mathematical matrix, and obtains the recognition model through the mathematical matrix; each column corresponds to the four-dimensional information of a key point;

A model algorithm module, which divides the actions and gestures in the mathematical matrix using a time window according to the recognition model, dynamically adjusts the actions and gestures after the time window division by setting an adjustment coefficient, and outputs a model result;

The model result scoring module evaluates the model as a whole according to the output result of the recognition model to obtain the Tai Chi scoring result.

In a third aspect, the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method when executing the computer program.

In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the method when executed by a processor.

Compared with the prior art, the present invention has the following advantages and beneficial effects:

The present invention utilizes time windows to divide actions and gestures in a mathematical matrix and determines key frames or feature points, thereby achieving accurate division of actions. By setting adjustment coefficients, the actions and gestures after time window division are dynamically adjusted, and the model results are output, which effectively improves the robustness of the model and ensures the accuracy of subsequent scores of the model.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

FIG1 is an overall flow chart of a time window-based Tai Chi scoring method according to an embodiment of the present invention;

FIG2 is a system structure diagram of a time window-based Tai Chi scoring method according to an embodiment of the present invention.

Detailed ways

In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the 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 cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

Example 1

Referring to FIG. 1 , which is a first embodiment of the present invention, the embodiment provides a Tai Chi scoring method based on a time window, comprising:

S1. Collect key points of Tai Chi movements in Tai Chi videos and define key points in each frame of the video.

Furthermore, each frame contains several key points, each of which contains coordinate information and gesture markers.

Furthermore, there are N key points, and the information of each key point is represented as a four-dimensional vector (x, y, t, g).

Among them, x and y are the coordinates of the key point in the Tai Chi video, t is the timestamp corresponding to the Tai Chi video, and g is the gesture marker.

It should be noted that the extraction and definition of key points help improve the accuracy of Tai Chi movements.

S2, taking the actions and gestures in the defined key points as a set of key data, constructing a mathematical matrix, and obtaining a recognition model through the mathematical matrix;

Furthermore, the actions and gestures defined in the key points are used as a set of key data to construct a mathematical matrix. Through the mathematical matrix, a recognition model is obtained, including:

Furthermore, we set the number of frames to M and express the matrix as , each row corresponds to the data of a time point, each column corresponds to the four-dimensional information of a key point, and the mathematical matrix X is constructed, which is expressed as:

;

It is understandable that, for the sake of ease of expression and processing, this embodiment expands this three-dimensional data set into a two-dimensional matrix. In this two-dimensional matrix, the information of all key points in the first frame is first arranged in sequence, then the second frame, and so on, until the last frame. Specifically, , , , They represent the horizontal coordinate, vertical coordinate and marked gesture of the first key point in the first frame, which are arranged continuously in a two-dimensional matrix; , , , Respectively represent the corresponding information of the second key point in the first frame, and are also arranged continuously in the matrix; and so on until , , , Indicates the information of the Nth key point in the first frame. Next, the key point information of the second frame , , , Start arranging until the information of the Nth key point of the last frame , , , In this way, each row of matrix data contains the coordinate information and gesture marks of all key points in a time point (i.e., a frame). In addition, the time of all key points in each frame is the same, i.e., associated with the frame itself. Through this expanded two-dimensional matrix, the present embodiment can construct a recognition model.

Furthermore, the input data X is passed to the shared feature extractor h, and the output h(X) of the shared feature extractor is used as the action recognition sub-network and gesture recognition subnetwork Finally, the output of model f is:

;

Each sub-network uses the features h(X) provided by the shared feature extractor to produce the recognition results of its own task, that is, and They are the recognition results of actions and gestures respectively.

It should be noted that when constructing the mathematical matrix X by inputting the number of frames, it is necessary to delete the continuous frames whose key point detection number does not reach the set threshold and the image frames whose key point detection number does not meet the set threshold, and adjust the timestamp t corresponding to the Tai Chi video.

Specifically, if the number of key point detections is not satisfied, that is, the x and y coordinates are missing, "NaN" is used to fill the corresponding missing points.

S3. According to the recognition model, the actions and gestures in the mathematical matrix are divided using the time window, and the actions and gestures after the time window division are dynamically adjusted by setting the adjustment coefficient, and the model results are output;

Furthermore, the time window sequence in the recognition model is:

;

The gesture sequence is:

;

Among them, each represents the i-th time window in the video, each Indicates the mth gesture.

Furthermore, the start frame of the gesture sequence The start frame of the corresponding time window sequence , starting from the frame after the start frame of the time window, the current frame of each frame of the gesture sequence is the start frame of the time window of the next frame, until the current frame of the gesture sequence is equal to the start frame of the time window, and the result is .

specific, Expressed as:

;

It should be noted that and The equal sign in is an assignment operation.

It should be noted that the current frame of the gesture sequence is equal to the start frame of the time window, which can realize real-time analysis and feedback of the gesture and can realize application optimization of the algorithm.

Furthermore, set the double adjustment coefficient and ;

Furthermore, through For time window series Adjust to obtain the change sequence corresponding to the gesture sequence ;

specific, Expressed as:

;

In the change sequence On this basis, the recognition model is adjusted to obtain:

;

in, Indicates that at time point t Adjusted time series values; is the time window series adjustment coefficient, For the change sequence Dynamic coefficient of represents the time period for identifying the model, The score of Tai Chi output by the recognition model.

It should be noted that setting the adjustment coefficient makes the model more flexible, which helps to improve the robustness of the model.

Furthermore, we define the model loss function and introduce binary variables Determine the model output results;

Furthermore, The output layer represents the action, The output layer representing the gesture, the recognition model output is expressed as:

;

Specifically, the loss function is divided into action loss function and gesture loss function;

Specifically, the action loss function Expressed as:

;

Specifically, the gesture loss function Expressed as:

;

Preferably, by setting the weight of the action part , the weight of the gesture part , and get the total loss value Expressed as:

;

further, and The value of is True or False; Expressed as binary cross entropy.

It should be noted that the use of binary cross entropy as the loss function is to perform category processing when actions and gestures are not equal, which helps to explain the model output results.

Furthermore, if the binary variable more than the , it is marked as 1, which is a trustworthy model; otherwise, it is marked as 0, which is an untrustworthy model, and returns to the construction of the mathematical matrix to rebuild the model.

S4. Based on the model output results, the model is evaluated as a whole to obtain the Tai Chi score results.

Based on the same idea as the time window-based Tai Chi scoring method in the above embodiment, the present invention also provides a time window-based Tai Chi scoring system, which can be used to execute the time window-based Tai Chi scoring method. For ease of explanation, the structural diagram of the embodiment of the time window-based Tai Chi scoring system only shows the parts related to the embodiment of the present invention. Those skilled in the art can understand that the illustrated structure does not constitute a limitation on the device, and may include more or fewer components than the illustrated one, or combine certain components, or arrange the components differently.

Furthermore, this embodiment also provides a Tai Chi scoring system based on a time window, comprising:

A video acquisition module, which acquires key points of Tai Chi movements in Tai Chi videos and defines key points in each frame of the video;

A model building module uses the actions and gestures in the defined key points as a set of key data to construct a mathematical matrix, and obtains a recognition model through the mathematical matrix;

A model algorithm module, which divides the actions and gestures in the mathematical matrix using a time window according to the recognition model, dynamically adjusts the actions and gestures after the time window division by setting an adjustment coefficient, and outputs a model result;

The model result scoring module evaluates the model as a whole according to the output result of the recognition model to obtain the Tai Chi scoring result.

This embodiment further provides a computer device, which is applicable to the Tai Chi scoring method based on a time window, and includes:

Memory and processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the Tai Chi scoring method based on time window as proposed in the above embodiment.

The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the time window-based Tai Chi scoring method proposed in the above embodiment is implemented.

The storage medium proposed in this embodiment and the data storage method proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

Example 2

Referring to Table 1 and Table 2, the second embodiment of the present invention provides a time window-based Tai Chi scoring method, including: verifying the accuracy of Tai Chi simulation of the method of the present invention by means of simulation experiments;

The loss function defined by the model , get the score of each Tai Chi action ,Each simulation experiment includes: 24 Tai Chi movements, each movement is scored ranging from 1 to 100 points.

The experiments were divided into two independent groups: the first group conducted 10 experiments, marked as experiments 1 to 10, without using the method of the present invention; in this group, the model stopped judging the subsequent actions when the score was less than 5 points (excluding 5 points); refer to Table 1;

Table 1 Action scores of the first set of 10 experiments

The second group conducted 10 experiments, marked as experiments 11 to 20, using the method of the present invention to ensure that the model does not interrupt the judgment of subsequent actions during the entire scoring process by dividing the time window; refer to Table 2;

Table 2 Action scores of the second group of 10 experiments

From Table 1 and Table 2, it can be found that the method of the present invention can monitor more actions in Table 2, and the action evaluation scores, that is, the accuracy of Tai Chi, are higher than the first 10 groups of experimental data in Table 1;

This shows that the method of the present invention can improve the accuracy of Tai Chi techniques and the continuous evaluation of movements, thereby improving the accuracy of detection for subsequent model training.

Example 3

Referring to Table 3, which is a third embodiment of the present invention, this embodiment provides a Tai Chi scoring method based on a time window, including:

Generate a dataset containing 100 image frames; each image frame contains 10 students, and each student independently and randomly generates 15 to 25 key points of Tai Chi movements in the image frames, which represent the number of key points of the human body detected by the model; according to the numbering order, these 100 image frames are divided into 10 groups for detection, and each group includes 10 image frames;

In the key point generation of each student, if the number of key points that meet the condition is greater than or equal to 18, the student is defined as a passing target, otherwise it is defined as a failing target; for each image frame, if the number of passing targets in the image frame is greater than or equal to 7, then the image frame is marked as a passing picture; otherwise, it is marked as a failing picture;

At the group level, if the number of qualified images in a group is greater than or equal to 7, then the group is a qualified group; otherwise, it is a failed group;

The key points and image frames in the above conditions are used as the training and , until the model outputs the results; refer to Table 3;

Table 3 Comparison of the output results after inputting key points and image frames into the model training

It can be seen from Table 3 that the number of failed targets, pictures and groups is less than or equal to 1/2 of their own values, indicating that the time window and gesture sequence of the present invention can process continuous image frames without large fluctuations, thereby increasing the stability and robustness of the model.

Example 4

2 , which is a fourth embodiment of the present invention, the embodiment provides a time window-based Tai Chi scoring system, including: a central server 100 , an edge server 200 , a camera 300 , a microphone 400 , a display device 500 , and a network 600 ;

The central server 100 includes,

Image storage module 11: used to store video images of Tai Chi students;

Model training module 12: inputting video images of Tai Chi students for training, and outputting the trained model;

The edge server 200 includes,

Data storage module 21: used to store the four-dimensional vector of key points generated by the video image;

Image analysis module 22: uses the model output by model training module 12 to analyze the video images of Tai Chi students collected by camera 300;

The camera 300 includes,

Video acquisition module 31: used to acquire video images of Tai Chi students;

The microphone 400 includes,

Information notification module 41: issues information notification according to the analysis result output by the image analysis module 22;

The display device 500 includes,

Information display module 51: displays evaluation results and improvement suggestions according to the analysis results output by the image analysis module 22;

The network 600 is used to connect the central server 100 , the edge server 200 , the camera 300 , the microphone 400 , and the display device 500 .

It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present application may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

The present application is described with reference to the flowcharts and/or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and/or box in the flowchart and/or block diagram, as well as the combination of the processes and/or boxes in the flowchart and/or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate a device for implementing the functions specified in one process or multiple processes in the flowchart and/or one box or multiple boxes in the block diagram.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device that implements the functions specified in one or more processes in the flowchart and/or one or more boxes in the block diagram.

These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes in the flowchart and/or one or more boxes in the block diagram.

Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims (8)

1. A Tai Chi scoring method based on a time window, characterized by comprising: Collect key points of Tai Chi movements in Tai Chi videos, and define key points in each frame of the video, wherein the key points include coordinate information and gesture marks; The coordinate information and gesture marks of the defined key points are used as a set of key data to construct a mathematical matrix, and a recognition model is obtained through the mathematical matrix; each column corresponds to the four-dimensional information of a key point; According to the recognition model, the actions and gestures in the mathematical matrix are divided by using a time window, the actions and gestures after the time window division are dynamically adjusted by setting an adjustment coefficient, and a model result is output; According to the output results of the recognition model, the model is evaluated as a whole to obtain a Tai Chi score result; The coordinate information and gesture marks of the defined key points are used as a set of key data to construct a mathematical matrix, and a recognition model is obtained through the mathematical matrix, including: When the number of frames is input into the mathematical matrix X, it is necessary to delete the continuous frames whose key point detection number does not reach the set threshold and the image frames whose key point detection number does not meet the set threshold, and adjust the timestamp t corresponding to the Tai Chi video; By setting the adjustment coefficient, the actions and gestures after the time window division are dynamically adjusted, and the model results are output, including: Setting dual adjustment coefficients and ; pass For time window series Adjust to obtain the change sequence corresponding to the gesture sequence ; In the change sequence On this basis, the recognition model is adjusted to obtain: ; in, Indicates that at time point t Adjusted time series values; is the time window series adjustment coefficient, For the change sequence Dynamic coefficient of represents the time period for identifying the model, The score of Tai Chi output by the recognition model; Inputting the X into the recognition model f, we get: ; in, and They are the recognition results of actions and gestures respectively. 2. The time window-based Tai Chi scoring method according to claim 1, characterized in that key points of Tai Chi movements in Tai Chi videos are collected and each frame number in the key points is defined, including: For each frame, there are several key points, each of which contains coordinate information and gesture markers; There are N key points, and the information of each key point is represented as a four-dimensional vector (x, y, t, g); Among them, x and y are the coordinates of the key point in the Tai Chi video, t is the timestamp corresponding to the Tai Chi video, and g is the gesture marker. 3. The time window-based Tai Chi scoring method according to claim 2, wherein the recognition model comprises: Set the number of frames to M and express the matrix as , construct the mathematical matrix X, expressed as: ; Among them, X represents a multidimensional matrix composed of M frames and N key points. , , , Respectively represent the abscissa, ordinate, time and marked gesture of the first key point in the first frame; , , , Respectively represent the abscissa, ordinate, time and marked gesture of the second key point in the first frame; , , , Respectively represent the horizontal coordinate, vertical coordinate, time and marked gesture of the Nth key point in the first frame; , , , Respectively represent the abscissa, ordinate, time and marked gesture of the first key point in the second frame; , , , Represent the horizontal coordinate, vertical coordinate, time and marked gesture of the Nth key point in the Mth frame respectively. 4. The time window-based Tai Chi scoring method according to claim 3, characterized in that, according to the recognition model, the actions and gestures in the mathematical matrix are divided using the time window, including: The time window sequence in the recognition model is: , the gesture sequence is , Among them, each represents the i-th time window in the video, each represents the mth gesture; The start frame of the gesture sequence The starting frame corresponding to the time window sequence , starting from the frame after the start frame of the time window, the current frame of each frame of the gesture sequence is the start frame of the time window of the next frame, until the current frame of the gesture sequence is equal to the start frame of the time window, and the time window sequence is obtained . 5. The time window-based Tai Chi scoring method according to claim 1, further comprising: Define the model loss function and introduce binary variables The output result of the recognition model is judged; the loss function includes an action loss function and a gesture loss function; If the binary variable more than the , it is marked as 1, which is a trustworthy model; otherwise, it is marked as 0, which is an untrustworthy model, and returns to the construction of the mathematical matrix to rebuild the model. 6. A time window-based Tai Chi scoring system, based on the time window-based Tai Chi scoring method according to any one of claims 1 to 5, characterized in that it comprises: A video acquisition module, which acquires key points of Tai Chi movements in Tai Chi videos and defines key points in each frame of the video, wherein the key points include coordinate information and gesture marks; The model building module uses the coordinate information and gesture marks of the defined key points as a set of key data to construct a mathematical matrix, and obtains the recognition model through the mathematical matrix; each column corresponds to the four-dimensional information of a key point; A model algorithm module, which divides the actions and gestures in the mathematical matrix using a time window according to the recognition model, dynamically adjusts the actions and gestures after the time window division by setting an adjustment coefficient, and outputs a model result; The model result scoring module evaluates the model as a whole according to the output result of the recognition model to obtain the Tai Chi scoring result. 7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of any one of the methods of claims 1 to 5 when executing the computer program. 8. A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.