Horizontal bar and parallel bar gymnastic action evaluation scoring method, device, equipment, medium and product

Abstract

The invention discloses a horizontal bar and parallel bar gymnastic action evaluation scoring method, device and equipment, a medium and a product, and relates to the technical field of sports data processing, and the method comprises the steps: processing a horizontal bar and parallel bar gymnastic action image based on a three-dimensional human body posture estimation algorithm, and obtaining human body key points; based on the human body key points, the target action sequence and the standard action sequence are compared and analyzed according to a DTW algorithm, and a preliminary distance metric value is obtained; based on the preliminary distance metric value, evaluating the target action sequence according to a feature threshold method, detecting angle change of joints in the target action sequence and position information of body parts, and deducting corresponding scores from feature quantities exceeding a preset threshold range to obtain a score deduction result; the characteristic quantity is a joint angle and a joint coordinate; and performing weighted fusion processing on the initial distance metric value and the score deduction result to obtain a comprehensive score of the target action sequence so as to comprehensively capture key features and detail information of the action.

Description

Technical Field

[0001] The present application relates to the technical field of sports data processing, and in particular to a method, device, equipment, medium and product for evaluating and scoring single and double bar gymnastics movements. Background Art

[0002] When evaluating the similarity of human motions based on human skeleton information, the basic idea is to compare the Euclidean distances between the key points of the two skeleton information in each frame, add them frame by frame, and use the sum as the similarity evaluation result of the two motion sequences. However, there are still serious problems in practical applications; when each athlete performs the same action, the speed and time of the action will vary greatly. For complex action combinations or continuous actions, it is difficult to effectively capture their key features and dynamic changes. Some actions may involve the coordinated movement of multiple body parts, and the temporal and spatial relationships between the parts are complex. This makes it impossible to compare the human skeleton key point information obtained from the two ends of the video frame by frame, which makes it very difficult to evaluate the subsequent action similarity. It is necessary to perform action segmentation or action recognition on the human skeleton key point information to understand the action features.

[0003] In view of the above problems, the present application proposes a method of dynamic time warping algorithm and feature threshold fusion to realize human motion similarity evaluation based on human skeleton. DTW algorithm (Dynamic Time Warping) is a classic and effective sequence matching algorithm, which has unique advantages in similarity measurement of time series data. It can find the optimal matching path between action sequences of different lengths and speeds, so as to realize quantitative evaluation of action similarity; DTW algorithm (Dynamic Time Warping) can ignore the difference in action time, focus on the morphology and trajectory characteristics of the action, so as to accurately compare the similarity between the action of the movement and the standard action model, and provide valuable training information for the coach. However, there are certain limitations in simply relying on DTW algorithm (Dynamic Time Warping) for action evaluation. Due to the complexity and diversity of the action, only considering sequence matching may not be able to fully capture the key features and detailed information of the action. Different actions often have specific key features or threshold ranges, which play a decisive role in accurately judging the quality and correctness of the action. Summary of the invention

[0004] The purpose of this application is to provide a method, device, equipment, medium and product for evaluating and scoring single and double bars gymnastics movements to solve the problem that sequence matching may not be able to fully capture the key features and detailed information of the movements.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In the first aspect, the present application provides a method for evaluating and scoring single and double bars gymnastics movements, specifically comprising:

[0007] Based on the 3D human posture estimation algorithm, the single and double bar gymnastics action images are processed to obtain the key points of the human body.

[0008] Based on the human body key points, the target action sequence is compared and analyzed with the standard action sequence according to the DTW algorithm to obtain a preliminary distance measurement value.

[0009] Based on the preliminary distance measurement value, the target action sequence is evaluated according to the feature threshold method, the angle changes of the joints and the position information of the body parts in the target action sequence are detected, and the feature quantities that exceed the preset threshold range are deducted with corresponding scores to obtain the deduction results; the feature quantities are joint angles and joint coordinates.

[0010] The preliminary distance measurement value and the deduction result are weightedly fused to obtain a comprehensive score of the target action sequence.

[0011] The single and double bar gymnastics movements were evaluated according to the composite score.

[0012] In the second aspect, the present application provides a single and double bar gymnastics movement evaluation and scoring device, specifically comprising:

[0013] The human body key point acquisition module processes the single and double bar gymnastics action images based on the three-dimensional human body posture estimation algorithm to obtain the human body key points.

[0014] The preliminary distance measurement value determination module compares and analyzes the target action sequence with the standard action sequence based on the human body key points according to the DTW algorithm to obtain a preliminary distance measurement value.

[0015] The deduction result determination module evaluates the target action sequence based on the preliminary distance measurement value according to the feature threshold method, detects the angle changes of the joints and the position information of the body parts in the target action sequence, and deducts the corresponding score from the feature quantities that exceed the preset threshold range to obtain the deduction result.

[0016] The comprehensive score determination module performs weighted fusion processing on the preliminary distance measurement value and the deduction result to obtain a comprehensive score of the target action sequence, and evaluates the single and double bar gymnastics movements according to the comprehensive score.

[0017] In a third aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the above-described single and double bar gymnastics movement evaluation and scoring methods.

[0018] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-mentioned single and double bar gymnastics movement evaluation and scoring methods.

[0019] In a fifth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of any of the above-mentioned single and double bar gymnastics movement evaluation and scoring methods.

[0020] According to the specific embodiments provided in this application, this application has the following technical effects:

[0021] The present application provides a method, device, equipment, medium and product for evaluating and scoring single and double bars gymnastics movements. The method processes the collected video data based on a three-dimensional human posture estimation algorithm to obtain human key points; based on the human key points, the target movement sequence is compared and analyzed with the standard movement sequence according to the DTW algorithm (Dynamic Time Warping) to obtain a preliminary distance measurement value, and the sequence matching is considered to accurately compare the similarity between the movement and the standard movement model; and based on the preliminary distance measurement value, the target movement sequence is evaluated according to the feature threshold method, the angle change of the joints and the position information of the body parts in the target movement sequence are detected, and the feature quantity exceeding the preset threshold range is deducted from the corresponding score to obtain the deduction result. The present application uses the feature threshold method to compare different movements in the target movement sequence with the corresponding threshold range respectively, remove the feature quantity exceeding the preset threshold range and deduct the corresponding score, so that the deduction result is more accurate; finally, based on the DTW algorithm (Dynamic Time Warping) The preliminary distance measurement result obtained by dynamic time warping (DTW) and the deduction result obtained based on the feature threshold method are weighted fused to obtain a comprehensive score of the target action sequence. This application not only takes sequence matching into consideration, but also performs feature threshold evaluation on different action features. It can not only judge the overall relevance of the action, but also conduct a detailed analysis on whether the key details in the action meet the standards, thereby fully capturing the key features and detail information of the action, and realizing a comprehensive and in-depth evaluation of all aspects of the action. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0023] Figure 1 A schematic diagram of a flow chart of a method for evaluating and scoring single and double bars gymnastics movements provided in an embodiment of the present application;

[0024] Figure 2 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0026] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0027] The present invention relates to the technical field of sports data processing, and in particular to a method for evaluating and scoring movements of single and double bars gymnastics. The method obtains a movement sequence of any athlete, obtains the athlete's three-dimensional human posture in gymnastics movements by a three-dimensional human posture estimation algorithm Faster-voxelpose algorithm, and simultaneously obtains three-dimensional coordinates of 25 key points of the athlete's body. The data features and sequence of the movement sequence are obtained by the above coordinates, and the difference between the athlete and the standard movement is calculated by a DTW algorithm (Dynamic Time Warping, dynamic time warping) and a movement evaluation algorithm of feature threshold fusion, so as to evaluate and score the athlete's single and double bars gymnastics movements.

[0028] like Figure 1 As shown, the embodiment of the present application provides a method for evaluating and scoring the movements of single and double bars gymnastics, which specifically includes:

[0029] S1: Based on the 3D human posture estimation algorithm, the single and double bar gymnastics action images are processed to obtain the key points of the human body.

[0030] S2: Based on the human body key points, the target action sequence is compared and analyzed with the standard action sequence according to the DTW algorithm to obtain a preliminary distance measurement value.

[0031] S3: Based on the preliminary distance measurement value, the target action sequence is evaluated according to the feature threshold method, the angle changes of the joints and the position information of the body parts in the target action sequence are detected, and the feature quantities exceeding the preset threshold range are deducted corresponding scores to obtain the deduction results; the feature quantities are joint angles and joint coordinates.

[0032] S4: Perform weighted fusion processing on the preliminary distance measurement value and the deduction result to obtain a comprehensive score of the target action sequence, and evaluate the single and double bar gymnastics movements according to the comprehensive score.

[0033] Further, in an exemplary embodiment, step S1 may be replaced by the following steps.

[0034] S101: extracting the three-dimensional human key point sequence information contained in each frame of the image in the single and double bars gymnastics movements, and determining the human key points; each frame of the image uses body25 key points to describe the structure and state information of the human skeleton; the body25 key points include the head, shoulders, elbows, wrists, hips, knees and ankles of the human body.

[0035] The overall process of the action evaluation algorithm using the DTW algorithm (Dynamic Time Warping) and feature threshold fusion is as follows. First, the collected video data is processed using the 3D human posture estimation algorithm Faster-voxelpose. The core of this processing process is to extract the 3D human key point sequence information contained in each frame of the video. Each frame uses body25 key points to describe the structure and state information of the human skeleton in detail. These key points cover the main joints of the human body and some body parts with key representation significance, such as the head, shoulders, elbows, wrists, hips, knees, ankles, etc. Together, they constitute a relatively complete and information-rich human skeleton model, which provides basic data support for subsequent action evaluation.

[0036] Further, in an exemplary embodiment, step S2 may be replaced by the following steps.

[0037] S201: constructing a grid matrix with the time series length of the target action sequence as rows and the time series length of the standard action sequence as columns.

[0038] S202: Based on the dynamic programming principle of the DTW algorithm, determine the distance between any two time series elements in the grid matrix.

[0039] S203: Determine the preliminary distance measurement value according to the distance between any two time series elements; the preliminary distance measurement value is the distance with the smallest cumulative difference from the starting point of the time series element to the end point of the time series element.

[0040] S204: Use formula Determine the distance between any two time series elements; where d(i,j) is the distance between any two time series elements; k is the feature dimension, and the range of k is 1 to m; the target action feature matrix F m(i, k) is the target action sequence; standard action feature matrix F′ m (j,k) is the standard action sequence.

[0041] S205: Use formula

[0042] Determine the cumulative distance; where d(i,j) represents the distance between the current two points; γ(i-1,j)+d(i,j) represents that when the standard action sequence is at the starting point and the target action sequence is at the i-th time point, the cumulative distance is the cumulative distance at the previous time point plus the distance between the current two points; γ(i,j-1)+d(i,j) represents that when the standard action sequence is at the starting point and the target action sequence is at the j-th time point, the cumulative distance is the cumulative distance at the previous time point plus the distance between the current two points; min(γ(i-1,j),γ(i,j-1),γ(i-1,j-1))+d(i,j) represents that when neither time series is at the starting point, the cumulative distance is the minimum value of the cumulative distance at the previous time point (which may be from the left, top, or upper left) plus the distance between the current two points.

[0043] S206: Filter the distance with the smallest cumulative difference from the starting point of the time series element to the end point of the time series element in the cumulative distances as a preliminary distance measurement value.

[0044] After the human body key point sequence is extracted, the DTW algorithm (Dynamic Time Warping) is used to evaluate the action, and the target action sequence (i.e. the action sequence to be evaluated) and the standard action sequence (pre-set standard action sequence) are carefully compared and analyzed according to the selected joint features. Specifically, by calculating the difference between the two sequences in each corresponding joint feature dimension, and based on the dynamic programming principle of the DTW algorithm (Dynamic Time Warping), the deformation on the time axis is comprehensively considered to find a matching path that minimizes the cumulative difference, thereby obtaining a preliminary distance measurement value of the two action sequences at the joint feature level.

[0045] In human motion assessment, the joint angles formed by human limbs and the physical positions of each joint are of great significance for the assessment of the overall motion. Choosing the right joint position and the right joint angle is crucial for the accuracy of motion assessment. In this experiment, the elbow angle, knee angle, angle between the two legs, foot position and hand position are selected as motion features.

[0046] Taking the left elbow joint as an example, assuming that the key point coordinates corresponding to the left shoulder of the human body in the current frame are v5 = (x5, y5, z5), the key point coordinates corresponding to the left elbow are v6 = (x6, y6, z6), and the key point coordinates corresponding to the left wrist of the human body are v7 = (x7, y7, z7). The calculation formula of the joint angle is as follows:

[0047]

[0048] in A vector from v6 to v5. is a vector pointing from v6 to v7, and θ is the angle between the vectors.

[0049] The Dynamic Time Warping (DTW) algorithm plays a vital role in the field of time series analysis, especially in processing time series data with time axis deformation or inconsistent sequence lengths. One of the most significant advantages of the DTW algorithm (Dynamic Time Warping) is that it can effectively handle the deformation of time series on the time axis. In practical applications, many time series data will be distorted in the time axis due to various factors. The DTW algorithm (Dynamic Time Warping) can adapt well to the stretching or compression of the time axis and find the best match between two time series by finding the optimal path through dynamic programming, while traditional methods based on fixed time alignment (such as simple Euclidean distance metric) may fail in this case.

[0050] Further, in an exemplary embodiment, step S3 may be replaced by the following step.

[0051] Assume that two time series records are X(i), where i = 1, 2, ..., m, and Y(j), where j = 1, 2, ..., n. Define a regular path W to represent the alignment or mapping of time series X(i) and Y(j), W = {w1(i, j), w2(i, j), ..., w k (i,j)}, k=1,2,...,p,p∈[max(m,n),m+n-1), where p represents the length of W. An m×n grid matrix is ​​constructed with X(i) sequence length m as rows and Y(j) sequence length n as columns. The DTW algorithm (Dynamic Time Warping) plans the optimal path from the starting point to the end point by finding the grid point with the shortest cumulative distance γ(i,j) obtained by adding d(i,j). The cumulative distance γ(i,j) of any point in the grid is calculated as follows:

[0052]

[0053] Where d(i,j) is the distance between the elements in the x(i) sequence and the elements in the y(j) sequence, γ(i,j) is the cumulative distance of d(i,j) iteratively calculated from (1,1) to the grid at (i,j), and min(γ(i-1,j),γ(i,j-1),γ(i-1,j-1)) means selecting the point with the smallest cumulative distance to continue the iterative calculation.

[0054] This study uses the target action feature matrix Fm to replace X(i) and the standard action feature matrix F'm to replace Y(j). Since the feature matrix is ​​composed of angle features and distance features, the calculation formula of d(i,j) is as follows:

[0055]

[0056] Further, in an exemplary embodiment, step S4 may be replaced by the following steps.

[0057] S301: Based on the joint angle range and body joint position requirements in the standard action sequence, the target action sequence is processed frame by frame to detect the angle changes of the joints and the position information of the body parts in the target action sequence.

[0058] S302: If the angle change of the angle-changed joint and the position information of the body part exceed the feature quantity within a preset threshold range, a corresponding score is deducted according to a deduction rule to obtain a deduction result.

[0059] This distance value reflects to some extent the degree of similarity deviation between the target action sequence and the standard action sequence in terms of overall form and rhythm, but it is only based on the dynamic matching of joint features, and has not yet fully considered the detailed normative requirements of the action. Next, a method based on feature thresholds is introduced to conduct a more in-depth and detailed evaluation of the action. According to the joint angle range and body joint position requirements unique to a specific action sequence, the target action sequence is processed frame by frame in a refined manner. In this process, the angle changes of each joint in the target action sequence and the position information of the body parts are detected in detail. Once the feature quantity is found to exceed the preset threshold range, the corresponding score is deducted according to the established deduction rules. This feature threshold-based evaluation method can effectively capture the deviation of the action in terms of detailed norms, thereby quantitatively evaluating the accuracy and normativeness of the action.

[0060] Further, in an exemplary embodiment, S4 may be replaced by the following steps.

[0061] S401: Perform weighted summation on the preliminary distance measurement value and the deduction result according to their respective weights to obtain a comprehensive score of the target action sequence.

[0062] S402: Evaluate the single and double bar gymnastics movements according to the comprehensive score.

[0063] The preliminary distance measurement results obtained based on the DTW algorithm (Dynamic Time Warping) and the deduction results obtained based on the feature threshold method are weighted and fused. According to the importance of the two methods in action evaluation and their respective advantages and limitations, reasonable weight coefficients are set for them respectively. By weighting and summing the two results according to their respective weights, the comprehensive score of the target action sequence is finally obtained, which intuitively reflects the similarity between the target action sequence and the standard action sequence. This weighted fusion method fully integrates the advantages of the DTW algorithm (Dynamic Time Warping) in processing the overall similarity of action sequences and the advantages of the feature threshold method in evaluating the standardization of action details, thereby achieving a more comprehensive, accurate and scientific evaluation of the action, and providing a reliable and in-depth quantitative basis for subsequent action analysis, training guidance and sports performance evaluation.

[0064] In gymnastics, a highly standardized and technical sport, the accuracy and standardization of movements play a decisive role in the performance evaluation and technical improvement of athletes. Gymnastics movements are diverse and complex, and each movement has its own unique motion trajectory, body coordination, and rhythm and rhythm requirements. It is difficult to fully and deeply capture the standardization of gymnastics movements in various dimensions by simply comparing the similarity of the characteristic indicators of two human skeleton sequences; for example, in the three movements on the horizontal bar, the athlete's body posture not only needs to be roughly consistent with the standard movement in terms of the overall movement, but also needs to strictly meet the standard requirements in terms of body angle, the opening between the two legs, and the arms and legs being straight when supporting; and in the three movements on the parallel bars, the angle formed by the body swinging back and forth is the key factor in measuring whether the movement is standard or not. The simple comparison of the similarity of skeleton sequence characteristics cannot fully consider these deep-level movement connotations and detailed requirements.

[0065] In view of this, in order to build a more scientific, comprehensive and accurate action evaluation system, we have communicated and discussed with senior gymnastics teachers and experienced coaches of a certain college. With many years of teaching practice experience and a deep understanding of the essentials of gymnastics techniques, teachers and coaches elaborated in detail the specific angle ranges that each joint of the human body should reach in different stages of movement in the horizontal bar three and parallel bars three gymnastics movements; these joint angle data were systematically sorted and analyzed and determined as the key characteristic thresholds in the action evaluation process. For example, in the single-leg swing into a support action of the horizontal bar three, the angle formed by the body with the ground when it is supported should be between 30 degrees and 60 degrees to ensure the rationality of muscle force and the standardization of the action. For example, in the split-leg sitting action of the parallel bars, the abduction angle of the hip joint and the bending angle of the knee joint must accurately meet the corresponding threshold standards in different forward steps. The abduction angle of the hip joint should reach a certain range when the legs are split, and the knee joint should be bent to facilitate the movement and conversion of the center of gravity of the body. Any deviation in the joint angle may cause the action to freeze or deform. By comparing and comprehensively judging the joint angle information in the actual gymnastics movement data of the athletes with these pre-set feature thresholds, it is possible to more effectively determine whether the gymnastics movements have truly met the standard requirements, thereby providing a solid and reliable quantitative basis and technical support for the evaluation of the training effect, improvement of technical movements and improvement of the competitive level of gymnasts. The specific thresholds selected for the three movements on the horizontal bar are shown in Table 1, and the specific thresholds selected for the three movements on the parallel bars are shown in Table 2.

[0066] Table 1 Specific thresholds for the three horizontal bar movements

[0067]

[0068]

[0069] Table 2 Specific thresholds selected for the three parallel bar movements

[0070]

[0071] The present application provides a single and double bar gymnastics movement evaluation and scoring device, which specifically includes:

[0072] The human body key point acquisition module processes the single and double bar gymnastics action images based on the three-dimensional human body posture estimation algorithm to obtain the human body key points.

[0073] The preliminary distance measurement value determination module compares and analyzes the target action sequence with the standard action sequence based on the human body key points according to the DTW algorithm to obtain a preliminary distance measurement value.

[0074] The deduction result determination module evaluates the target action sequence based on the preliminary distance measurement value according to the feature threshold method, detects the angle changes of the joints and the position information of the body parts in the target action sequence, and deducts the corresponding score from the feature quantities that exceed the preset threshold range to obtain the deduction result.

[0075] The comprehensive score determination module performs weighted fusion processing on the preliminary distance measurement value and the deduction result to obtain a comprehensive score of the target action sequence, and evaluates the single and double bar gymnastics movements according to the comprehensive score.

[0076] The embodiment of the present application provides a computer device, which may be a server or a terminal, and its internal structure diagram may be as follows: Figure 2 As shown, the computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface; wherein the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. 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. The database of the computer device is used to store video tag processing data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a video tag processing method is implemented.

[0077] An embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the single and double bar gymnastics movement evaluation and scoring method described above is implemented.

[0078] 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.

[0079] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for evaluating and scoring single and double bar gymnastics movements, characterized in that: The single and double bars evaluation and scoring method comprises: Based on the 3D human posture estimation algorithm, the single and double bar gymnastics action images are processed to obtain the key points of the human body; Based on the human body key points, the target action sequence is compared and analyzed with the standard action sequence according to the DTW algorithm to obtain a preliminary distance measurement value; Based on the preliminary distance measurement value, the target action sequence is evaluated according to the feature threshold method, the angle changes of the joints and the position information of the body parts in the target action sequence are detected, and the feature quantities exceeding the preset threshold range are deducted corresponding points to obtain the deduction result; the feature quantities are the joint angles and joint coordinates; Performing weighted fusion processing on the preliminary distance measurement value and the deduction result to obtain a comprehensive score of the target action sequence; The single and double bar gymnastics movements were evaluated according to the composite score.

2. The method for evaluating and scoring single and double bars gymnastics movements according to claim 1, characterized in that: The method of processing the single and double bar gymnastics action images based on the three-dimensional human posture estimation algorithm to obtain the key points of the human body specifically includes: The three-dimensional human key point sequence information contained in each frame of the single and double bars gymnastics movements is extracted to determine the human key points; each frame of the image uses body25 key points to describe the structure and state information of the human skeleton; the body25 key points include the head, shoulders, elbows, wrists, hips, knees and ankles of the human body.

3. The method for evaluating and scoring single and double bars gymnastics movements according to claim 1, characterized in that: Based on the human body key points, the target action sequence is compared and analyzed with the standard action sequence according to the DTW algorithm to obtain a preliminary distance measurement value, which specifically includes: Constructing a grid matrix with the time series length of the target action sequence as rows and the time series length of the standard action sequence as columns; Based on the dynamic programming principle of the DTW algorithm, the distance between any two time series elements in the grid matrix is ​​determined; The preliminary distance measurement value is determined according to the distance between any two time series elements; the preliminary distance measurement value is the distance with the smallest cumulative difference from the starting point of the time series element to the end point of the time series element.

4. The method for evaluating and scoring single and double bars gymnastics movements according to claim 3, characterized in that: The preliminary distance measurement value is the distance with the smallest cumulative difference, specifically including: Using the formula Determine the distance between any two time series elements; where d(i,j) is the distance between any two time series elements; the target action feature matrix F m (i,k) and F′ m (j, k) are the elements in the target action feature matrix and the standard action feature matrix respectively; k is the feature dimension, and the range of k is 1 to m; using the formula Determine the cumulative distance; where d(i,j) represents the distance between the current two points; γ(i-1,j)+d(i,j) represents that when the standard action sequence is at the starting point and the target action sequence is at the i-th time point, the cumulative distance is the cumulative distance at the previous time point plus the distance between the current two points; γ(i,j-1)+d(i,j) represents that when the standard action sequence is at the starting point and the target action sequence is at the j-th time point, the cumulative distance is the cumulative distance at the previous time point plus the distance between the current two points; min(γ(i-1,j),γ(i,j-1),γ(i-1,j-1))+d(i,j) represents that when both time series are not at the starting point, the cumulative distance is the minimum value of the cumulative distance at the previous time point (which may be from the left, top, or top left) plus the distance between the current two points; The distance with the smallest cumulative difference from the starting point of the time series element to the end point of the time series element is selected as the preliminary distance measurement value.

5. The method for evaluating and scoring the single and double bars gymnastics movements according to claim 1, wherein the target movement sequence is evaluated based on the preliminary distance metric value according to the feature threshold method, the angle changes of the joints and the position information of the body parts in the target movement sequence are detected, and the feature quantities exceeding the preset threshold range are deducted corresponding scores to obtain the deduction results, specifically including: Based on the joint angle range and body joint position requirements in the standard action sequence, the target action sequence is processed frame by frame to detect the angle changes of the joints and the position information of the body parts in the target action sequence; If the angle change of the angle-changed joint and the position information of the body part exceed the characteristic quantity within the preset threshold range, the corresponding score is deducted according to the deduction rules to obtain the deduction result.

6. The method for evaluating and scoring single and double bars gymnastics movements according to claim 1, characterized in that: The preliminary distance measurement value and the deduction result are weighted and fused to obtain a comprehensive score of the target action sequence, specifically including: Performing a weighted summation of the preliminary distance measurement value and the deduction result according to their respective weights to obtain a comprehensive score of the target action sequence; The single and double bar gymnastics movements were evaluated according to the composite score.

7. A single and double bar gymnastics movement evaluation and scoring device, characterized in that: The single and double bars gymnastics movement evaluation and scoring device comprises: The human body key point acquisition module processes the single and double bar gymnastics action images based on the 3D human body posture estimation algorithm to obtain the human body key points; A preliminary distance measurement value determination module, based on the human body key points, compares and analyzes the target action sequence with the standard action sequence according to the DTW algorithm to obtain a preliminary distance measurement value; A deduction result determination module, based on the preliminary distance measurement value, evaluates the target action sequence according to a feature threshold method, detects angle changes of joints and position information of body parts in the target action sequence, and deducts corresponding scores from feature quantities exceeding a preset threshold range to obtain a deduction result; The comprehensive score determination module performs weighted fusion processing on the preliminary distance measurement value and the deduction result to obtain a comprehensive score of the target action sequence, and evaluates the single and double bar gymnastics movements according to the comprehensive score.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the single and double bar gymnastics movement evaluation and scoring method according to any one of claims 1 to 6.

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 single and double bar gymnastics movement evaluation and scoring method described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the single and double bar gymnastics movement evaluation and scoring method described in any one of claims 1 to 6 is implemented.

Images