Taijiquan coherence scoring method based on dynamic sliding window and related equipment

Abstract

The invention provides a dynamic sliding window-based shadowboxing coherence scoring method and related equipment. The method comprises the following steps of: obtaining a shadowboxing video; 3D skeleton points and space coordinates are extracted from the shadowboxing video; according to the shadowboxing style to which the frame picture belongs, distributing a weight level for each 3D skeleton point; according to the weight levels of the 3D skeleton points, assigning a sliding window range for each 3D skeleton point; judging whether the space coordinates of the 3D skeleton points are changed or not in a specified sliding window range, and if not, judging that the action is lagging; respectively counting the number of lagging times of each 3D skeleton point, and calculating a greeting style coherence score of each shadowboxing greeting style according to the number of lagging times; according to the invention, different sliding window ranges are allocated according to the weight levels of the 3D skeleton points, so that the computing resources of the system and the accuracy of detection are balanced; the jamming times of each 3D skeleton point are counted, and the coherence score of each Tai Chi style is calculated according to the jamming times, so that the fluency degree of the action can be evaluated more comprehensively.

Description

Technical Field

[0001] The present invention relates to the technical field of motion recognition, and more specifically, to a Tai Chi continuity scoring method based on a dynamic sliding window and related equipment. Background Art

[0002] Tai Chi is an important part of the traditional culture of the Chinese nation and occupies an important position in traditional Chinese martial arts. The existing Tai Chi learning methods are mainly through imitating Tai Chi movements in books, videos and other media or learning under the guidance of professional personnel. Basic Tai Chi movements and postures can be learned through books or videos, but the mastery of the movement essentials and the understanding of boxing techniques are subject to certain limitations. Conventional offline teaching is mainly in the form of one-to-many, and the personal teaching experience and skills of teachers vary greatly. During the teaching process, it is impossible to pay attention to the changes in everyone's movements and postures at any time.

[0003] In addition, in professional Tai Chi and traditional martial arts competitions, the scoring is mainly based on whether the completed movements are correct, coherent, coordinated, and the coordination between the movements and the music. However, due to personal experience and other factors, the judges will inevitably have scoring biases. Therefore, in the face of the problems in Tai Chi in public participation and competition, it is necessary to design an intelligent movement quality analysis method and give trainers reasonable feedback to improve the posture of the movements.

[0004] The Chinese invention patent (CN202410424786.3) proposes a Tai Chi scoring method, system, device and medium based on a time window, which collects the key points of Tai Chi movements in Tai Chi videos and defines the key points in each frame of the video; the movements 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; according to the recognition model, the movements and gestures in the mathematical matrix are divided using a time window, and the movements and gestures after the time window division are dynamically adjusted by setting an adjustment coefficient, and the model results are output; the present invention divides the movements and gestures in the mathematical matrix using a time window, determines the key frames, and thus realizes the precise division of the movements; by setting an adjustment coefficient, the movements 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.

[0005] However, the technical solution in this patent still has the following defects: the factors considered by the Tai Chi scoring algorithm are too general and single, and the focus area cannot be changed according to the Tai Chi moves. Summary of the invention

[0006] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a Tai Chi continuity scoring method and related equipment based on a dynamic sliding window, so as to overcome the shortcomings of the existing Tai Chi scoring algorithm that the factors considered are too single and the focus area cannot be changed according to the Tai Chi moves.

[0007] The above technical objectives of the present invention are achieved through the following technical solutions: A Tai Chi continuity scoring method based on a dynamic sliding window, comprising: Obtain a Tai Chi video; extract 3D skeleton points and spatial coordinates corresponding to the 3D skeleton points from each frame of the Tai Chi video; According to the Tai Chi moves to which the frame belongs, a weight level is assigned to each 3D bone point in each frame; According to the weight level corresponding to the 3D skeleton point, a sliding window range is specified for each 3D skeleton point; and it is determined whether the change amplitude of the spatial coordinates of the 3D skeleton point corresponding to the sliding window range exceeds a predetermined change amplitude threshold within the specified sliding window range, and if not, it is determined that the action is stuck; The number of freezes of each 3D skeleton point is counted respectively, and the move continuity score of each Tai Chi move is calculated according to the number of freezes.

[0008] In one embodiment, the extracting of 3D skeleton points and the spatial coordinates corresponding to the 3D skeleton points from each frame of the Tai Chi video specifically includes: extracting 2D skeleton points and the plane coordinates corresponding to the 2D skeleton points from each frame of the Tai Chi video; based on the temporal relationship between the frames and the spatial relationship between the plane coordinates, converting the 2D skeleton points into a 2D skeleton point sequence; converting the 2D skeleton point sequence into initial 3D skeleton points; and performing enhancement operations on the initial 3D skeleton points to generate spatial coordinates.

[0009] In one embodiment, extracting 2D skeleton points and the plane coordinates corresponding to the 2D skeleton points from each frame of the Tai Chi video specifically includes: using a convolutional neural network to extract 2D skeleton points and the plane coordinates corresponding to the 2D skeleton points from each frame.

[0010] In one embodiment, the converting of the 2D skeleton points into a 2D skeleton point sequence based on the temporal relationship between the frame images and the spatial relationship between the plane coordinates specifically includes: using a Transformer model to capture the temporal relationship between all 2D skeleton points and the spatial relationship between the plane coordinates, and converting the 2D skeleton points into a 2D skeleton point sequence containing spatiotemporal continuity.

[0011] In one embodiment, converting the 2D skeleton point sequence into initial 3D skeleton points specifically includes: inputting the 2D skeleton point sequence into a multi-layer perceptron to generate initial 3D skeleton points.

[0012] In one embodiment, the enhancing operation on the initial 3D skeleton points to generate spatial coordinates specifically includes: utilizing a multi-head self-attention mechanism to enhance the initial 3D skeleton points to generate 3D skeleton points and corresponding spatial coordinates.

[0013] In one embodiment, the method of assigning a weight level to each 3D skeleton point in each frame according to the Tai Chi moves to which the frame belongs specifically includes: using an LSTM model to identify the Tai Chi moves in each frame; based on the Tai Chi moves to which each frame belongs, and in accordance with preset expert rating rules, marking a weight level for each 3D skeleton point in each frame, the weight levels specifically include primary weight, secondary weight and tertiary weight.

[0014] In one embodiment, the sliding window range is specified for each 3D skeleton point according to the weight level corresponding to the 3D skeleton point, specifically including: the sliding window range includes a sliding window size and a sliding window step; when the 3D skeleton point is a first-level weight, the sliding window size is 60 frames, and the sliding window step is 12 frames; when the 3D skeleton point is a second-level weight, the sliding window size is 30 frames, and the sliding window step is 6 frames; when the 3D skeleton point is a third-level weight, the sliding window size is 15 frames, and the sliding window step is 3 frames.

[0015] In one embodiment, the method of counting the number of freezes of each 3D skeleton point respectively and calculating the move continuity score of each Tai Chi move according to the number of freezes specifically includes: calculating the skeleton point continuity score according to the number of freezes of the 3D skeleton point and the weight level of the 3D skeleton point; calculating the move continuity score of each Tai Chi move based on the skeleton point continuity score and the affiliation between the 3D skeleton point and the Tai Chi move.

[0016] In one embodiment, the calculation of the bone point continuity score according to the number of freezes of the 3D bone points and the weight level of the 3D bone points specifically includes: calculating the number of sliding windows corresponding to each 3D bone point based on the time length of the Tai Chi moves and the weight level of each 3D bone point; assigning weight scores to the 3D bone points according to the weight levels of the 3D bone points; wherein, when the 3D bone point has a first-level weight, the weight score is 1 point, when the 3D bone point has a second-level weight, the weight score is 2 points, and when the 3D bone point has a third-level weight, the weight score is 3 points; based on the number of freezes and the number of sliding windows, calculating the weight score discount corresponding to each 3D bone point; and using the weight score discount and weight score of the 3D bone point to calculate the bone point continuity score accordingly.

[0017] In one embodiment, the calculation of the move continuity score of each Tai Chi move based on the skeleton point continuity score and the affiliation between the 3D skeleton point and the Tai Chi moves specifically includes calculating the move continuity score using the following formula:

[0018] in, Indicates The rating of the Tai Chi moves; Indicates The total score of the Tai Chi moves; Indicates The sum of the first-level weighted bone point continuity scores in the Tai Chi moves; Indicates The sum of the coherence scores of all secondary weighted bone points in the Tai Chi moves; Indicates The sum of the continuity scores of all three-level weighted bone points in the Tai Chi moves; Indicates The total number of 3D bone points with first-level weight in the Tai Chi moves; Represents the total number of 3D bone points with secondary weights in the i-th Tai Chi move; Indicates The total number of 3D bone points with third-level weights in the Tai Chi moves.

[0019] In one embodiment, the method further includes, based on all the move coherence scores, calculating the Tai Chi coherence score based on the dynamic sliding window, specifically including: using the following formula to calculate the Tai Chi coherence score based on the dynamic sliding window:

[0020] in, Indicates the move continuity score, Represents the Tai Chi continuity score based on the dynamic sliding window.

[0021] A Tai Chi continuity scoring device based on a dynamic sliding window, comprising: An extraction unit is used to obtain a Tai Chi video; extract 3D skeleton points and spatial coordinates corresponding to the 3D skeleton points from each frame of the Tai Chi video; An allocating unit, for allocating a weight level to each 3D skeleton point in each frame according to the Tai Chi moves to which the frame belongs; A judging unit is used to specify a sliding window range for each 3D skeleton point according to the weight level corresponding to the 3D skeleton point; and judge whether the change amplitude of the spatial coordinates of the 3D skeleton point corresponding to the sliding window range exceeds a predetermined change amplitude threshold within the specified sliding window range, and if not, it is judged that the action is stuck; The scoring unit is used to count the number of freezes of each 3D skeleton point respectively, and calculate the move continuity score of each Tai Chi move according to the number of freezes.

[0022] A computer-readable storage medium stores a computer program, which implements the steps of the above method when executed by a processor.

[0023] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0024] In summary, the present invention has the following beneficial effects: a Tai Chi continuity scoring method based on dynamic sliding window, comprising: obtaining Tai Chi video; extracting 3D skeleton points and spatial coordinates corresponding to the 3D skeleton points from each frame of the Tai Chi video; assigning a weight level to each 3D skeleton point in each frame according to the Tai Chi moves to which the frame belongs; specifying a sliding window range for each 3D skeleton point according to the weight level corresponding to the 3D skeleton point; and judging whether the change amplitude of the spatial coordinates of the 3D skeleton points corresponding to the sliding window range exceeds a predetermined change amplitude threshold within the specified sliding window range, and if not, determining that the action is stuck; counting the number of stuck times for each 3D skeleton point respectively, and calculating the move coherence score of each Tai Chi move according to the number of stuck times; using the method of the present invention, different sliding window ranges are assigned to the 3D skeleton points according to their weight levels, effectively balancing the computing resources of the system and the accuracy of detection, especially at important skeleton points in action evaluation, and more subtle incoherence can be captured. By counting the number of freezes at each 3D skeleton point and calculating the continuity score of each Tai Chi move based on the number of freezes, we can more comprehensively evaluate the smoothness of the movements. This is not limited to overall evaluation, but can also be accurate to the continuity performance of specific skeleton points. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a flow chart of a Tai Chi continuity scoring method based on a dynamic sliding window of the present invention; Figure 2 is a structural diagram of a Tai Chi continuity scoring device based on a dynamic sliding window in an embodiment of the present invention; Figure 3 1 is a diagram showing the internal structure of a computer device in an embodiment of the present invention.

[0026] In the figure: 1. Extraction unit; 2. Allocation unit; 3. Judgment unit; 4. Scoring unit. DETAILED DESCRIPTION

[0027] In order to make the purpose, 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 accompanying drawings. Several embodiments of the present invention are shown in the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein.

[0028] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal connection of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances. The terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0029] In the present invention, unless otherwise clearly specified and limited, the first feature being "above" or "below" the second feature may include the first and second features being in direct contact, or may include the first and second features being in contact not directly but through another feature between them. Moreover, the first feature being "above", "above" and "above" the second feature includes the first feature being directly above and obliquely above the second feature, or simply indicates that the first feature is higher in level than the second feature. The first feature being "below", "below" and "below" the second feature includes the first feature being directly below and obliquely below the second feature, or simply indicates that the first feature is lower in level than the second feature. The terms "vertical", "horizontal", "left", "right", "above", "below" and similar expressions are for illustrative purposes only, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0030] Embodiment 1 In order to solve the problems existing in the prior art, the present invention provides a Tai Chi continuity scoring method based on a dynamic sliding window, such as Figure 1 As shown, including: S1, obtaining a Tai Chi video; extracting 3D skeleton points and spatial coordinates corresponding to the 3D skeleton points from each frame of the Tai Chi video; S2, assigning a weight level to each 3D skeleton point in each frame according to the Tai Chi moves to which the frame belongs; S3, specifying a sliding window range for each 3D skeleton point according to the weight level corresponding to the 3D skeleton point; and determining whether the change amplitude of the spatial coordinates of the 3D skeleton point corresponding to the sliding window range exceeds a predetermined change amplitude threshold within the specified sliding window range, and if not, determining that the action is stuck; S4. Count the number of freezes for each 3D skeleton point respectively, and calculate the move continuity score of each Tai Chi move according to the number of freezes.

[0031] In actual use, in order to be able to score each Tai Chi action in the video separately, and change the focus of attention according to the different Tai Chi moves, this application proposes a Tai Chi continuity scoring method, first obtain the Tai Chi video, that is, use a camera to shoot the Tai Chi performance video of the person being scored, and extract 3D skeleton points and corresponding spatial coordinates from the Tai Chi performance video. In practical applications, it is necessary to restrict the format of the Tai Chi video, the frame rate of the video, etc., so that data processing can be based on the same standard to ensure the fairness of the scoring. In this embodiment, the format of the video is set to MP4 format, the frame rate is 30 frames per second, and the resolution is 1080p. In each frame, there are 17 skeleton points of the human body, including the top of the head (Head Top), neck (Neck), left shoulder (Left Shoulder), right shoulder (Right Shoulder), left elbow (Left Elbow), right elbow (Right Elbow), left wrist (Left Wrist), right wrist (Right Wrist), left hip (Left Hip), right hip (Right Hip), left knee (Left Knee), right knee (Right Knee), left ankle (Left Ankle), right ankle (Right Ankle), chest / torso (Chest / Torso), left toe (Left Toe), right toe (Right Toe). By identifying these skeleton points in the video and calculating their spatial coordinates , , …, , where xyz represents the spatial coordinates of the bone points, and the Indicates The number in the superscript indicates the In the frame One of the bone points.

[0032] During the recognition process, it is also necessary to identify the Tai Chi moves in the frame. In the performance, not all people's skeletal points have large movements. Instead, when the end of the limb moves with a large amplitude, the skeletal points of the torso are driven to move with a small amplitude, or in the case of hand moves, the foot moves with a small amplitude. Therefore, when judging whether the movement is coherent, it is necessary to focus on one or more skeletal points based on different moves. Therefore, it is necessary to assign a weight level to each skeletal point based on the amplitude of the movement of the move. The higher the weight level of the skeletal point, the greater the amplitude of movement of the skeletal point, and it needs to be paid special attention. For example, in the move of Cloud Hand, the amplitude of movement of the wrist position is the largest, so the skeletal points of the two wrists can be assigned to the third level of weight, and the amplitude of movement of the foot is small and not obvious, so it can be assigned to the first level of weight. The allocation of skeletal points in other parts can be adaptively allocated according to the actual situation or the method recommended by expert scoring.

[0033] During the recognition process, the judgment of coherence can be calculated by the number of times of reverse recognition jams and then deducted from the preset basic score to achieve the score of Tai Chi coherence. According to the weight of the 3D bone point, the sliding window technology is used in the video to identify and judge the movement state of the bone point. When the movement of the bone point within a certain sliding window range does not exceed the preset change amplitude threshold, it means that the bone point has jammed, and then the corresponding score needs to be deducted. Here, the number of jams corresponding to the bone point is calculated. Therefore, when calculating the move score, the number of jams of all bone points needs to be counted, and finally a move coherence score is output. Suppose that in the "Cloud Hand" move, the elbow (first-level weight point) has 5 jams, the knee (second-level weight point) has 2 jams, and the wrist (third-level weight point) has no jams. The system calculates according to the weight and number of jams of each bone point. Assume that the final coherence score of "Cloud Hand" is 85 points (out of 100 points). For the specific calculation process, please refer to the subsequent content.

[0034] In summary, this application forms a technical solution by combining multiple steps such as accurate 3D posture recognition, flexible skeleton point weight allocation mechanism, dynamic sliding window mechanism, action jam detection, and fine-grained coherence scoring, which is more objective and accurate in evaluating Tai Chi coherence and can provide strong technical support for practitioners and reviewers. It can also solve the defect that the judgment of the coherence of moves in the prior art is too general and cannot be based on the detailed weight allocation of each move, which affects the accuracy of the scoring.

[0035] In one embodiment, the extracting of 3D skeleton points and the spatial coordinates corresponding to the 3D skeleton points from each frame of the Tai Chi video specifically includes: extracting 2D skeleton points and the plane coordinates corresponding to the 2D skeleton points from each frame of the Tai Chi video; based on the temporal relationship between the frames and the spatial relationship between the plane coordinates, converting the 2D skeleton points into a 2D skeleton point sequence; converting the 2D skeleton point sequence into initial 3D skeleton points; and performing enhancement operations on the initial 3D skeleton points to generate spatial coordinates.

[0036] In this application, the plane coordinates of the skeleton points in the video are first identified, and then the plane coordinates are converted into a skeleton point sequence, and finally the 2D skeleton point sequence is converted into a 3D space coordinate, wherein the 2D skeleton point sequence is a time series data, which contains the position changes of the joint points over a period of time (multiple frames). By modeling these sequences, the temporal motion trajectory and posture changes of the human body can be captured, and the spatiotemporal relationship between the joint points is captured through the spatiotemporal model. These relationships help the model understand the coherence of the human body's motion between multiple frames, thereby providing a basis for subsequent 3D posture inference, and the 2D skeleton point sequence can be used as a basis for correcting posture estimation errors. Even if the 2D joint point detection result of a certain frame is wrong, based on the sequence data of the previous and next frames, the model can use the continuity in time to smooth or correct these errors, so as to ensure the coherence and accuracy of the final generated 3D posture.

[0037] In one embodiment, extracting 2D skeleton points and the plane coordinates corresponding to the 2D skeleton points from each frame of the Tai Chi video specifically includes: using a convolutional neural network to extract 2D skeleton points and the plane coordinates corresponding to the 2D skeleton points from each frame.

[0038] Specifically, in this embodiment, a convolutional neural network is used to extract the plane coordinates of the skeleton points from each frame. First, each frame is preprocessed separately, and the image size is adjusted to make it more in line with the input requirements of the convolutional neural network; a pre-trained convolutional neural network CNN is used as the skeleton. This network is composed of multiple convolutional layers, pooling layers, activation functions (such as ReLU), and fully connected layers, and can learn features in the image. The results of the CNN output specifically include 17 heat maps, each of which corresponds to the position probability distribution of a skeleton point. On each joint point heat map, the position with the highest value is selected as the two-dimensional coordinate of the joint point. The final output result of CNN is a two-dimensional coordinate list, which represents the 2D plane position of all skeleton points in the frame image. By repeatedly processing each video frame, the combined result is the processing of the Tai Chi video.

[0039] In one embodiment, the converting of the 2D skeleton points into a 2D skeleton point sequence based on the temporal relationship between the frame images and the spatial relationship between the plane coordinates specifically includes: using a Transformer model to capture the temporal relationship between all 2D skeleton points and the spatial relationship between the plane coordinates, and converting the 2D skeleton points into a 2D skeleton point sequence containing spatiotemporal continuity.

[0040] The Transformer model mainly includes the following modules: Position module, which provides position information for the input 2D skeleton point sequence through position encoding. For each frame of the skeleton point sequence, the corresponding frame index position encoding is added so that the model can capture the information of the time dimension. Self-attention mechanism: It can capture the relationship between each skeleton point in the input sequence. Here, it can simultaneously capture the temporal relationship (the relationship between the same skeleton point in different frames) and the spatial relationship (the relationship between skeleton points in the same frame). By inputting the flattened 2D skeleton point sequence into the self-attention module, the model can calculate the dependency of each skeleton point with other skeleton points in the temporal and spatial dimensions. Multi-head attention mechanism: Using multiple parallel attention heads, each head can learn different temporal and spatial feature representations. Through the multi-head attention mechanism, the model can capture the complex spatiotemporal relationship between skeleton points at different scales. Each attention head finds different patterns and dependencies in the spatiotemporal relationship, and then combines these patterns to obtain a richer feature representation. For example, a Tai Chi video clip containing 100 frames is used, with 2D coordinates of 17 skeleton points in each frame. First, the 2D coordinates of the skeleton points of each frame are extracted to generate a shape of The input matrix of 34 is because the plane coordinates corresponding to the 17 skeleton points have two horizontal and vertical coordinates. The temporal position information is added to each frame of the skeleton point sequence through position encoding, and then input into the Transformer model. The multi-head self-attention mechanism is used to capture the spatiotemporal relationship of all skeleton points. After passing through the multi-layer Transformer encoder, a 2D skeleton point sequence containing spatiotemporal continuity features is output. The output generated by the final Transformer model can be directly used for subsequent action coherence analysis or posture evaluation.

[0041] In one embodiment, converting the 2D skeleton point sequence into initial 3D skeleton points specifically includes: inputting the 2D skeleton point sequence into a multi-layer perceptron to generate initial 3D skeleton points.

[0042] Specifically, the multilayer perceptron contains several hidden layers, each with multiple neurons. The hidden layer learns the mapping relationship from 2D skeleton points to 3D skeleton points through linear transformation and nonlinear activation function. The final output layer maps the results of the hidden layer to 3D space, and the output of each skeleton point is 3D. Therefore, the dimension of the output layer will be 17*3=51, representing 17 bone points in each frame. coordinate.

[0043] In one embodiment, the enhancing operation on the initial 3D skeleton points to generate spatial coordinates specifically includes: utilizing a multi-head self-attention mechanism to enhance the initial 3D skeleton points to generate 3D skeleton points and corresponding spatial coordinates.

[0044] In this embodiment, the multi-head self-attention mechanism uses multiple parallel attention heads, each of which focuses on different skeleton point relationships and can learn different spatiotemporal patterns. Each head independently calculates the above self-attention weights and generates a corresponding output vector. The output of each attention head will be connected in series and then mapped back to the original dimension through a linear transformation. In practical applications, if 8 attention heads are used, the output dimension will be 8 times the output dimension of each head. The output vector of each skeleton point is obtained by weighted summation. The value vectors of each skeleton point are weighted and summed according to the attention weights to generate new skeleton point features, which are further used to adjust the initial 3D coordinates.

[0045] In one embodiment, the method of assigning a weight level to each 3D skeleton point in each frame according to the Tai Chi moves to which the frame belongs specifically includes: using an LSTM model to identify the Tai Chi moves in each frame; based on the Tai Chi moves to which each frame belongs, and in accordance with preset expert rating rules, marking a weight level for each 3D skeleton point in each frame, the weight levels specifically include primary weight, secondary weight and tertiary weight.

[0046] Specifically, since it is not possible to directly mark each frame in the video individually, it is necessary to automatically identify the actions made in each frame based on the LSTM model. In this embodiment, the frame rate of the video is 24 frames per second. When a Tai Chi move lasts for 5-10 seconds, the total number of frames corresponding to a Tai Chi move is 120-240 frames; then in the frames covered by this Tai Chi move, the weight corresponding to each 3D skeleton point should be consistent. It should be noted that the first-level weight is the lightest weight, and the third-level weight is the heaviest weight. For example, in the "cloud hand" move, more attention is paid to hand movements, so the skeleton points of the wrist and elbow are marked as the third-level weight, and the skeleton points of the ankle are marked as the first-level weight. The weight level corresponding to each skeleton point is determined by pre-allocation. It can be evaluated by expert review in each move according to the importance and the corresponding weight is assigned so that the same weight standard can be used in the subsequent scoring process.

[0047] In one embodiment, the sliding window range is specified for each 3D skeleton point according to the weight level corresponding to the 3D skeleton point, specifically including: the sliding window range includes a sliding window size and a sliding window step; when the 3D skeleton point is a first-level weight, the sliding window size is 60 frames, and the sliding window step is 12 frames; when the 3D skeleton point is a second-level weight, the sliding window size is 30 frames, and the sliding window step is 6 frames; when the 3D skeleton point is a third-level weight, the sliding window size is 15 frames, and the sliding window step is 3 frames.

[0048] Specifically, when the weight level of the 3D bone point is higher, the corresponding movement amplitude of the 3D bone point is larger, that is to say, the model needs to pay more attention to the movement continuity of this bone point. Therefore, the higher the weight level, the smaller the weight sliding window size and the smaller the step size of the sliding window, so that the bone points that need to be focused on can be evaluated more carefully. For example: in the "Cloud Hand" move, the athlete's action time is 10 seconds, then the corresponding number of video frames is 240 frames; for the wrist bone point weight level is the third level weight, the sliding window size is 15 frames, the sliding window step is 3 frames, and the number of sliding windows for the wrist should be

[0049] in, Indicates the total number of frames of the "Cloud Hands" move. Indicates the number of sliding windows corresponding to the three-level weight, that is, the number of times the skeleton points of the three-level weight need to be monitored. In the "Cloud Hand" move, the number of detections of all three-level weights is For the ankle bone point weight level of level 1, the sliding window size is 60 frames, the sliding window step is 12 frames, and the number of sliding windows for the ankle should be:

[0050] in, Indicates the total number of frames of the "Cloud Hands" move. Indicates the number of sliding windows corresponding to the first-level weight, that is, the number of times the first-level weight skeleton points need to be monitored. In the "Cloud Hand" move, the number of detections for all first-level weights is .

[0051] Based on the above content, it can be known that in the actual recognition process, the higher the weight level, the more attention should be paid to the joint point. Therefore, the smaller the size of the sliding window, the smaller the corresponding time interval, and the stricter the detection result. In a short period of time, if the movement of a certain bone point does not exceed the predetermined threshold, it means that there is a jam at the joint point. For some first-level weights, it means that the bone point is not taken seriously in the Tai Chi moves, and the movement of the bone point can be detected in a longer time interval to avoid mistakenly identifying unnecessary movement as jam.

[0052] In one embodiment, the method of counting the number of freezes of each 3D skeleton point respectively and calculating the move continuity score of each Tai Chi move according to the number of freezes specifically includes: calculating the skeleton point continuity score according to the number of freezes of the 3D skeleton point and the weight level of the 3D skeleton point; calculating the move continuity score of each Tai Chi move based on the skeleton point continuity score and the affiliation between the 3D skeleton point and the Tai Chi move.

[0053] In one embodiment, the calculation of the bone point continuity score according to the number of freezes of the 3D bone points and the weight level of the 3D bone points specifically includes: calculating the number of sliding windows corresponding to each 3D bone point based on the time length of the Tai Chi moves and the weight level of each 3D bone point; assigning weight scores to the 3D bone points according to the weight levels of the 3D bone points; wherein, when the 3D bone point has a first-level weight, the weight score is 1 point, when the 3D bone point has a second-level weight, the weight score is 2 points, and when the 3D bone point has a third-level weight, the weight score is 3 points; based on the number of freezes and the number of sliding windows, calculating the weight score discount corresponding to each 3D bone point; and using the weight score discount and weight score of the 3D bone point to calculate the bone point continuity score accordingly.

[0054] Specifically, in actual use, in the "Cloud Hands" move, the athlete's action time is 10 seconds, so the corresponding number of video frames is 240 frames; then the calculation of the sliding window times for the three-level weighted skeleton points should be:

[0055] That is 75 times; assuming that 5 of them are judged as jams, then the proportion of 5 jams to all the judgment times should be , in order to highlight the proportional relationship between the jams and the total number This application also calculates the proportion of the number of freezes It is divided into three levels. ; ; .according to The interval to which it belongs, with the minimum value of the interval as the weight score discount Based on the above content, the weight score discount and weight score of the 3D skeleton point are used to calculate the skeleton point coherence score, which is calculated using the following formula: ;in, Indicates the score of a certain c-level weighted bone point, that is, Represents a first-level weighted bone point; however, there are 17 bone points in each frame of the video, and there are usually multiple first-level bone points, so , , There can be multiple ones, and you only need to calculate them separately during the calculation process. Represents the weight score of the c-level weight, that is, , , . Represents the weight score discount, In the case of ;exist In the case of ;exist In the case of According to the above calculation process, in the previous example, in the Cloud Hands move, the left wrist action of the third-level weighted bone point was stuck 5 times, so the continuity score of the left wrist bone point should be = 3. For another example, if the right ankle of the first-level weighted bone point is stuck 6 times, then the number of sliding windows is calculated first.

[0056] Then according to the ratio of the number of sliding windows to the number of freezes ,for ,because ,therefore ; The final bone point continuity score of the right ankle is:

[0057] In one embodiment, the calculation of the move continuity score of each Tai Chi move based on the skeleton point continuity score and the affiliation between the 3D skeleton point and the Tai Chi moves specifically includes calculating the move continuity score using the following formula:

[0058] in, Indicates The rating of the Tai Chi moves; Indicates The total score of the Tai Chi moves; Indicates The sum of the first-level weighted bone point continuity scores in the Tai Chi moves; Indicates The sum of the coherence scores of all secondary weighted bone points in the Tai Chi moves; Indicates The sum of the continuity scores of all three-level weighted bone points in the Tai Chi moves; Indicates The total number of 3D bone points with first-level weight in the Tai Chi moves; Represents the total number of 3D bone points with secondary weights in the i-th Tai Chi move; Indicates The total number of 3D bone points with third-level weights in the Tai Chi moves.

[0059] In this application, there are 24 Tai Chi moves, so For example, in the Cloud Hands move, there are 240 frames, each of which has 3 first-level weighted bone points, 5 second-level weighted bone points, and 9 third-level weighted bone points. In the entire Cloud Hands move video, there are 3*240=720 first-level weighted bone points with continuity scores, so The corresponding value is 720, and It is the sum of the continuity scores of these 720 first-level weighted bone points. Similarly, there are 5*240=1200 second-level weighted bone points in the Cloud Hands move, so The corresponding value is 1200, and It is the sum of the continuity scores of these 1200 second-level weighted bone points. Similarly, there are 9*240=2160 third-level weighted bone points in the Cloud Hands technique, so The corresponding value is 2160, and It is the sum of the continuity scores of these 2160 third-level weighted bone points.

[0060] In order to independently score the continuity of each Tai Chi move, this application distributes the scores of each Tai Chi move equally based on the percentage system. Tai Chi has a total of 24 moves, so the full score of each Tai Chi move should be 100 / 24=4.16, so .

[0061] In one embodiment, the method further includes, based on all the move coherence scores, calculating the Tai Chi coherence score based on the dynamic sliding window, specifically including: using the following formula to calculate the Tai Chi coherence score based on the dynamic sliding window:

[0062] in, Indicates the move continuity score, Represents the Tai Chi continuity score based on the dynamic sliding window.

[0063] In order to make an overall evaluation of a set of Tai Chi, the continuity score of each move is added together to obtain the Tai Chi continuity score based on the dynamic sliding window.

[0064] In summary, the Tai Chi coherence scoring method based on dynamic sliding window provided by the present application has accurate 3D posture recognition, flexible skeleton point weight allocation mechanism, dynamic sliding window mechanism, action jamming detection, and fine-grained coherence scoring compared with the prior art; according to the Tai Chi moves to which each frame belongs, a weight level is assigned to each 3D skeleton point. This weight mechanism can flexibly respond to the characteristics of different moves and highlight the most critical skeleton points in each move. By distinguishing the weight levels, the system can focus on the key parts of each move more accurately, and improve the scientificity and rationality of the scoring. According to the weight level of the 3D skeleton point, a different sliding window range is assigned to it. The weight-based sliding window mechanism makes the detection of key skeleton points more detailed and accurate, while the detection of secondary skeleton points is more relaxed. This sliding window method can dynamically adjust the accuracy of the detection, effectively balancing the system's computing resources and the accuracy of the detection, especially at the important skeleton points in the action evaluation, which can better capture subtle incoherence.

[0065] Embodiment 2 See also Figure 2 A Tai Chi continuity scoring device based on a dynamic sliding window, the Tai Chi continuity scoring device based on a dynamic sliding window comprising: Extraction unit 1, used for acquiring Tai Chi video; extracting 3D skeleton points and spatial coordinates corresponding to the 3D skeleton points from each frame of the Tai Chi video; The allocation unit 2 is used to allocate a weight level to each 3D skeleton point in each frame according to the Tai Chi moves to which the frame belongs; The judging unit 3 is used to specify a sliding window range for each 3D skeleton point according to the weight level corresponding to the 3D skeleton point; and judge whether the change amplitude of the spatial coordinates of the 3D skeleton point corresponding to the sliding window range exceeds a predetermined change amplitude threshold within the specified sliding window range, and if not, it is judged that the action is stuck; The scoring unit 4 is used to count the number of freezes of each 3D skeleton point respectively, and calculate the move continuity score of each Tai Chi move according to the number of freezes.

[0066] For the specific definition of the Tai Chi continuity scoring device based on dynamic sliding window, please refer to the definition of the Tai Chi continuity scoring method based on dynamic sliding window above, which will not be repeated here. Each module in the Tai Chi continuity scoring device based on dynamic sliding window can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0067] Those skilled in the art will understand that Figure 2The structure shown in the figure is only a block diagram of a part of the structure related to the present application scheme, and does not constitute a limitation of the present application scheme. The specific Tai Chi continuity scoring device based on dynamic sliding window may include more or less components than those shown in the figure, or combine certain components, or have a different component arrangement. Embodiment 3 A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the Tai Chi continuity scoring method based on a dynamic sliding window as described in Example 1.

[0068] Embodiment 4 In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a network interface and a database connected through 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, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. When the computer program is executed by the processor, a Tai Chi continuity scoring method based on a dynamic sliding window is implemented.

[0069] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0070] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the following steps when executing the computer program, including: obtaining a Tai Chi video; extracting 3D skeleton points and spatial coordinates corresponding to the 3D skeleton points from each frame of the Tai Chi video; assigning a weight level to each 3D skeleton point in each frame according to the Tai Chi moves to which the frame belongs; specifying a sliding window range for each 3D skeleton point according to the weight level corresponding to the 3D skeleton point; and determining whether the change amplitude of the spatial coordinates of the 3D skeleton points corresponding to the sliding window range exceeds a predetermined change amplitude threshold within the specified sliding window range, and if not, determining that a motion freeze occurs; and counting the number of freezes of each 3D skeleton point respectively, and calculating a move coherence score for each Tai Chi move according to the number of freezes.

[0071] In one embodiment, the extracting of 3D skeleton points and the spatial coordinates corresponding to the 3D skeleton points from each frame of the Tai Chi video specifically includes: extracting 2D skeleton points and the plane coordinates corresponding to the 2D skeleton points from each frame of the Tai Chi video; based on the temporal relationship between the frames and the spatial relationship between the plane coordinates, converting the 2D skeleton points into a 2D skeleton point sequence; converting the 2D skeleton point sequence into initial 3D skeleton points; and performing enhancement operations on the initial 3D skeleton points to generate spatial coordinates.

[0072] In one embodiment, extracting 2D skeleton points and the plane coordinates corresponding to the 2D skeleton points from each frame of the Tai Chi video specifically includes: using a convolutional neural network to extract 2D skeleton points and the plane coordinates corresponding to the 2D skeleton points from each frame.

[0073] In one embodiment, the converting of the 2D skeleton points into a 2D skeleton point sequence based on the temporal relationship between the frame images and the spatial relationship between the plane coordinates specifically includes: using a Transformer model to capture the temporal relationship between all 2D skeleton points and the spatial relationship between the plane coordinates, and converting the 2D skeleton points into a 2D skeleton point sequence containing spatiotemporal continuity.

[0074] In one embodiment, converting the 2D skeleton point sequence into initial 3D skeleton points specifically includes: inputting the 2D skeleton point sequence into a multi-layer perceptron to generate initial 3D skeleton points.

[0075] In one embodiment, the enhancing operation on the initial 3D skeleton points to generate spatial coordinates specifically includes: utilizing a multi-head self-attention mechanism to enhance the initial 3D skeleton points to generate 3D skeleton points and corresponding spatial coordinates.

[0076] In one embodiment, the method of assigning a weight level to each 3D skeleton point in each frame according to the Tai Chi moves to which the frame belongs specifically includes: using an LSTM model to identify the Tai Chi moves in each frame; based on the Tai Chi moves to which each frame belongs, and in accordance with preset expert rating rules, marking a weight level for each 3D skeleton point in each frame, the weight levels specifically include primary weight, secondary weight and tertiary weight.

[0077] In one embodiment, the sliding window range is specified for each 3D skeleton point according to the weight level corresponding to the 3D skeleton point, specifically including: the sliding window range includes a sliding window size and a sliding window step; when the 3D skeleton point is a first-level weight, the sliding window size is 60 frames, and the sliding window step is 12 frames; when the 3D skeleton point is a second-level weight, the sliding window size is 30 frames, and the sliding window step is 6 frames; when the 3D skeleton point is a third-level weight, the sliding window size is 15 frames, and the sliding window step is 3 frames.

[0078] In one embodiment, the method of counting the number of freezes of each 3D skeleton point respectively and calculating the move continuity score of each Tai Chi move according to the number of freezes specifically includes: calculating the skeleton point continuity score according to the number of freezes of the 3D skeleton point and the weight level of the 3D skeleton point; calculating the move continuity score of each Tai Chi move based on the skeleton point continuity score and the affiliation between the 3D skeleton point and the Tai Chi move.

[0079] In one embodiment, the calculation of the bone point continuity score according to the number of freezes of the 3D bone points and the weight level of the 3D bone points specifically includes: calculating the number of sliding windows corresponding to each 3D bone point based on the time length of the Tai Chi moves and the weight level of each 3D bone point; assigning weight scores to the 3D bone points according to the weight levels of the 3D bone points; wherein, when the 3D bone point has a first-level weight, the weight score is 1 point, when the 3D bone point has a second-level weight, the weight score is 2 points, and when the 3D bone point has a third-level weight, the weight score is 3 points; based on the number of freezes and the number of sliding windows, calculating the weight score discount corresponding to each 3D bone point; and using the weight score discount and weight score of the 3D bone point to calculate the bone point continuity score accordingly.

[0080] In one embodiment, the calculation of the move continuity score of each Tai Chi move based on the skeleton point continuity score and the affiliation between the 3D skeleton point and the Tai Chi moves specifically includes calculating the move continuity score using the following formula:

[0081] in, Indicates The rating of the Tai Chi moves; Indicates The total score of the Tai Chi moves; Indicates The sum of the first-level weighted bone point continuity scores in the Tai Chi moves; Indicates The sum of the coherence scores of all secondary weighted bone points in the Tai Chi moves; Indicates The sum of the continuity scores of all three-level weighted bone points in the Tai Chi moves; Indicates The total number of 3D bone points with first-level weight in the Tai Chi moves; Represents the total number of 3D bone points with secondary weights in the i-th Tai Chi move; Indicates The total number of 3D bone points with third-level weights in the Tai Chi moves.

[0082] In one embodiment, the method further includes, based on all the move coherence scores, calculating the Tai Chi coherence score based on the dynamic sliding window, specifically including: using the following formula to calculate the Tai Chi coherence score based on the dynamic sliding window:

[0083] in, Indicates the move continuity score, Represents the Tai Chi continuity score based on the dynamic sliding window.

[0084] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0085] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0086] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A Tai Chi continuity scoring method based on dynamic sliding window, characterized in that: include: Get Tai Chi videos; Extracting 3D skeleton points and spatial coordinates corresponding to the 3D skeleton points from each frame of the Tai Chi video; According to the Tai Chi moves to which the frame belongs, a weight level is assigned to each 3D bone point in each frame; According to the weight level corresponding to the 3D skeleton point, a sliding window range is specified for each 3D skeleton point; and it is determined whether the change amplitude of the spatial coordinates of the 3D skeleton point corresponding to the sliding window range exceeds a predetermined change amplitude threshold within the specified sliding window range, and if not, it is determined that the action is stuck; The number of freezes of each 3D skeleton point is counted respectively, and the move continuity score of each Tai Chi move is calculated according to the number of freezes.

2. The Tai Chi continuity scoring method based on dynamic sliding window according to claim 1 is characterized in that: The step of extracting 3D skeleton points and spatial coordinates corresponding to the 3D skeleton points from each frame of the Tai Chi video specifically includes: Extracting 2D skeleton points and plane coordinates corresponding to the 2D skeleton points from each frame of the Tai Chi video; Based on the temporal relationship between the frames and the spatial relationship between the plane coordinates, the 2D skeleton points are converted into a 2D skeleton point sequence; Convert the 2D skeleton point sequence into initial 3D skeleton points; An enhancement operation is performed on the initial 3D skeleton points to generate spatial coordinates.

3. The Tai Chi continuity scoring method based on dynamic sliding window according to claim 1 is characterized in that: The step of assigning a weight level to each 3D skeleton point in each frame according to the Tai Chi moves to which the frame belongs includes: Use the LSTM model to identify Tai Chi moves in each frame; Based on the Tai Chi moves to which each frame belongs, according to the preset expert rating rules, a weight level is marked for each 3D bone point in each frame, and the weight level specifically includes primary weight, secondary weight and tertiary weight.

4. The Tai Chi continuity scoring method based on dynamic sliding window according to claim 3 is characterized in that: The step of specifying a sliding window range for each 3D skeleton point according to the weight level corresponding to the 3D skeleton point specifically includes: The sliding window range includes the sliding window size and the sliding window step length; When the 3D skeleton points are of first-level weight, the sliding window size is 60 frames and the sliding window step is 12 frames; When the 3D skeleton points are secondary weighted, the sliding window size is 30 frames and the sliding window step is 6 frames; When the 3D skeleton points are three-level weighted, the sliding window size is 15 frames and the sliding window step is 3 frames.

5. The method for scoring Tai Chi continuity based on dynamic sliding window according to claim 4, characterized in that: The method of counting the number of freezes of each 3D skeleton point and calculating the move coherence score of each Tai Chi move according to the number of freezes specifically includes: The coherence score of the 3D skeleton points is calculated according to the number of freezes of the 3D skeleton points and the weight level of the 3D skeleton points; Based on the skeleton point continuity score and the affiliation between the 3D skeleton points and the Tai Chi moves, the move continuity score of each Tai Chi move is calculated.

6. The method for scoring Tai Chi continuity based on dynamic sliding window according to claim 5, characterized in that: The bone point coherence score is calculated according to the number of freezes of the 3D bone point and the weight level of the 3D bone point, specifically including: Based on the duration of Tai Chi moves and the weight level of each 3D skeleton point, the number of sliding windows corresponding to each 3D skeleton point is calculated; A weight score is assigned to the 3D skeleton point according to the weight level of the 3D skeleton point; wherein, when the 3D skeleton point is of the first-level weight, the weight score is 1 point, when the 3D skeleton point is of the second-level weight, the weight score is 2 points, and when the 3D skeleton point is of the third-level weight, the weight score is 3 points; Based on the number of freezes and the number of sliding windows, a weight score discount corresponding to each 3D skeleton point is calculated; The weight score discount and weight score of the 3D skeleton point are used to calculate the skeleton point continuity score.

7. The method for scoring Tai Chi continuity based on dynamic sliding window according to claim 6, characterized in that: The calculation of the move continuity score of each Tai Chi move based on the skeleton point continuity score and the affiliation between the 3D skeleton point and the Tai Chi moves specifically includes calculating the move continuity score using the following formula: in, Indicates The rating of the Tai Chi moves; Indicates The total score of the Tai Chi moves; Indicates The sum of the first-level weighted bone point continuity scores in the Tai Chi moves; Indicates The sum of the coherence scores of all secondary weighted bone points in the Tai Chi moves; Indicates The sum of the continuity scores of all three-level weighted bone points in the Tai Chi moves; Indicates The total number of 3D bone points with first-level weight in the Tai Chi moves; Represents the total number of 3D bone points with secondary weights in the i-th Tai Chi move; Indicates The total number of 3D bone points with third-level weights in the Tai Chi moves.

8. A Tai Chi continuity scoring device based on dynamic sliding window, characterized in that: The Tai Chi continuity scoring device based on dynamic sliding window comprises: An extraction unit is used to obtain a Tai Chi video; extract 3D skeleton points and spatial coordinates corresponding to the 3D skeleton points from each frame of the Tai Chi video; An allocating unit, for allocating a weight level to each 3D skeleton point in each frame according to the Tai Chi moves to which the frame belongs; A judging unit is used to specify a sliding window range for each 3D skeleton point according to the weight level corresponding to the 3D skeleton point; and judge whether the change amplitude of the spatial coordinates of the 3D skeleton point corresponding to the sliding window range exceeds a predetermined change amplitude threshold within the specified sliding window range, and if not, it is judged that the action is stuck; The scoring unit is used to count the number of freezes of each 3D skeleton point respectively, and calculate the move continuity score of each Tai Chi move according to the number of freezes.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the dynamic sliding window-based Tai Chi continuity scoring method as described in any one of claims 1 to 7 is implemented.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the dynamic sliding window-based Tai Chi continuity scoring method as described in any one of claims 1 to 7 is implemented.

Images