Automated performance recording and standardized action evaluation method and system for campus sports project test analysis

Abstract

The present invention discloses an automated score recording and standardized action evaluation method and system for campus sports project test analysis, relates to the technical field of sports project test analysis, and includes the following steps: configuring an end-side image acquisition and processing device and a cloud-side image processing device. First, the present invention automatically records the scores of the tested persons by analyzing the position information and trajectory information of the athletes in the physical test video clips, as well as the position information of the sports equipment; secondly, the present invention analyzes the athletes' movement postures, evaluates their standardization and normativeness according to the key points of sports technology, and provides personalized technical improvement suggestions for athletes. The present invention can not only automatically calculate and record scores, but also deeply analyze the details of the movements. By comparing with standard movements, it can identify the shortcomings in the athletes' movements and provide scientific improvement plans, which will help athletes better understand the technical requirements and thus improve their sports performance.

Description

Technical Field

[0001] The present invention relates to the technical field of sports project testing and analysis, and in particular to an automated performance recording and standardized action evaluation method and system for campus sports project testing and analysis. Background Art

[0002] The traditional physical education teaching evaluation system can no longer meet current needs. The traditional evaluation system mainly evaluates students' athletic ability, but it is difficult to make detailed and specific evaluations of students' individual differences. Therefore, it is necessary to introduce a more scientific and comprehensive evaluation method to improve the physical education teaching evaluation system and promote the personalized development of physical education.

[0003] In the prior art, manual measurement, data collection based on depth cameras, and data collection based on RGB cameras are often used for test analysis. However, the manual measurement method has subjective errors and is difficult to achieve digital recording. The method based on depth camera data collection has certain difficulties in identifying targets of similar depths, is difficult to accurately identify, and has high costs. The method based on RGB cameras is more dependent on image clarity and algorithm accuracy, consumes large computing resources, and is difficult to implement on ordinary equipment. Therefore, how to combine the data collection methods of depth cameras and RGB cameras, utilize their respective advantages, and improve the accuracy of test analysis is the problem we need to solve. To this end, this paper proposes an automated score recording and standardized action evaluation method and system for campus sports project test analysis. Summary of the invention

[0004] The purpose of the present invention is to provide an automated score recording and standardized action evaluation method and system for campus sports project testing and analysis, so as to solve the problems raised in the above-mentioned background technology.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] First, the automated performance recording and standardized action evaluation method for campus sports project test analysis includes the following steps:

[0007] Step 1: Configure the end-side image acquisition and processing device and the cloud-side image processing device, and start the end-side device and the cloud-side device respectively. The end-side device includes a depth camera, an RGB camera, and a tablet device, and the cloud-side device is a cloud-side server.

[0008] Step 2: Use a depth camera to collect campus physical test videos, obtain a human body depth map through human body depth estimation, perform 3D space modeling based on the depth information, perform preliminary calculations on the physical test scores, and obtain information related to human body movement;

[0009] Step 3: Display preliminary results on the client device for on-site personnel to view immediately.

[0010] Step 4: Determine the physical test status based on the human body depth information, determine the key time period, use the RGB camera to collect video clips of the key time period, and upload the preliminary scores and RGB video clips generated by the client device to the cloud server.

[0011] Step 5: In the cloud-side server, the uploaded video clips are processed using an image processing algorithm based on RGB images and a deep learning model to calculate the physical test scores and evaluate the movement posture;

[0012] Step 6: Based on the data processed by the cloud-side server, a customized physical test report generation model is used to generate a physical test report and return it to the relevant personnel.

[0013] A further improvement of the technical solution of the present invention is that in step 1, the process of configuring the end-side image acquisition and processing device and the cloud-side image processing device is as follows:

[0014] Step 101: configure the end-side image acquisition and processing device, wherein the end-side image acquisition and processing device includes a depth camera, an RGB camera, and a tablet device. Place the end-side device on a tripod at a specific location next to the test area, adjust the viewing angle and focal length of the camera so that the image collected by the camera can cover the entire test area, and connect the depth camera to the tablet device via a wired method. The RGB camera can be a built-in camera of the tablet device or an external device supporting the depth camera.

[0015] Step 102: Install corresponding image acquisition and processing software on the tablet device, configure the software, and receive data from the depth camera and the RGB camera. The camera and the tablet device are the end-side devices. Turn on the camera and the tablet device to check the end-side devices.

[0016] Step 103, configuring a cloud-side image processing device, wherein the cloud-side image processing device is a cloud-side server, and the cloud-side server is a cloud-side device, and installing and configuring an operating system and software environment of the cloud-side server;

[0017] Step 104: deploy image processing and deep learning models on the cloud-side server, and configure a network connection between the cloud-side server and the end-side device to receive data from the end-side device;

[0018] Step 105, respectively start the terminal side device and the cloud side device, check whether the communication between the devices is normal, use the tablet device and the camera in conjunction with the image acquisition and processing software to capture test data, use the cloud side server to receive the test data of the terminal side device, and monitor the data upload process.

[0019] A further improvement of the technical solution of the present invention is that in step 2, the process of acquiring the information related to human body movement is as follows:

[0020] Step 201, using a depth camera to collect campus physical measurement videos, and analyzing the depth image of the test area captured by the campus physical measurement video, and performing noise reduction and illumination compensation preprocessing operations on the captured depth image of the test area;

[0021] Step 202, using the depth image data, detecting a human body region from the depth image through an image processing algorithm, and extracting depth information of the human body region;

[0022] Step 203, based on the extracted human body area depth information, convert the key pixel points in the two-dimensional depth map into points in three-dimensional space to form point cloud data, and construct a three-dimensional space model of the key points of the human body through a surface reconstruction algorithm;

[0023] Step 204, based on the human body key point data in the 3D space and in combination with the rules of the physical test items, the trajectory of the human body movement is analyzed and the preliminary score is calculated.

[0024] A further improvement of the technical solution of the present invention is that in step 4, the process of collecting the video clips in the key time period is:

[0025] Step 401, analyzing the extracted human body area depth information, calculating the key movements and states of the human body, and determining the physical measurement state of the measured person;

[0026] Step 402, detecting the key action and state, determining the current moment as the beginning of the key time period, marking it, and determining the length of the key time period;

[0027] Step 403, based on the determined key time period, using an RGB camera to collect video of the key time period to obtain an RGB video clip;

[0028] Step 404: On the terminal device, a preliminary physical test score is calculated based on the data captured by the depth camera and the preset physical test rules, and the preliminary physical test score and the RGB video clip are packaged and uploaded to the cloud server.

[0029] A further improvement of the technical solution of the present invention is that in step 5, the process of calculating the physical test results and evaluating the action posture is as follows:

[0030] Step 501: In the cloud-side server, the uploaded RGB video clip is decoded into image frames, and an image enhancement algorithm is applied to process the image frames decoded from the RGB video clip to improve image quality;

[0031] Step 502, using a human body detection model based on deep learning to detect human body targets in each frame of image, and using a human body skeleton key point model based on deep learning to extract human body skeleton joint points;

[0032] Step 503, using the object detection model to detect the sports equipment in the image, and segmenting and classifying the equipment to extract its key features;

[0033] Step 504, combining the image position information of the human body and the sports equipment, the key point information of the human skeleton, and the actual size information of the human body and the sports equipment, to determine whether the person being measured is in a valid state and a key action stage of the physical test;

[0034] Step 505, according to the rules of the physical test items, the physical test scores are calculated using the extracted information, and the angles of specific joints of the human body are calculated according to the key points of the human skeleton, and the weighted action evaluation index is obtained by analysis to evaluate the accuracy and standardization of the action.

[0035] A further improvement of the technical solution of the present invention is that the calculation process of the weighted action evaluation index is:

[0036] According to the standard rules of physical test items, the action evaluation function is defined. By using this function, the action evaluation value can be obtained by calculating the angles of the joints of key parts of the human body.

[0037] From the video clips obtained by the RGB camera and the depth camera, computer vision technology is used to extract the location information of key points and analyze the standardization and regularity of the movements;

[0038] According to the action evaluation function and key point position information, the action evaluation value of each actual action is estimated, and a single action evaluation index is calculated. Finally, the weighted action evaluation index of all key actions is calculated through the weight matrix.

[0039] The human body position information and weighted motion evaluation index of key actions are combined to analyze the trajectory and posture of human movement and evaluate the standardization and normativeness of the overall movement.

[0040] A further improvement of the technical solution of the present invention is that in step 6, the expression of the single action evaluation index is:

[0041]

[0042]

[0043] in, For the The evaluation index of each action, For the The actual detected action of the action, For the The critical effective action of an action, For the The critical full score action of the action, is the action evaluation function, and the obtained value is the action evaluation value. and For , , Two bone vectors of three joints, where the joints can be any joints of the human body, including the head, neck, shoulder, elbow, hand, hip, knee, ankle, heel, and toe. Less than the effective threshold When the action evaluation value is Greater than the full score threshold When the evaluation index is 1, the action evaluation value Between the two thresholds, the evaluation index varies linearly between 0 and 1.

[0044] A further improvement of the technical solution of the present invention is that the expression of the weighted action evaluation index is:

[0045]

[0046]

[0047]

[0048] in, is the weighted action evaluation index, is the number of key actions, For the The weight of each action can be determined according to the importance of the action in different physical test items.

[0049] In step 6, the process of generating the physical test report is as follows:

[0050] Step 601, summarizing all relevant data processed in the cloud-side server, including physical test scores and key action information, and performing data standardization conversion;

[0051] Step 602, analyzing the physical test results, and comparing the key action information with the preset standards to determine whether the action meets the standard requirements, and generating action standardization evaluation results, including action evaluation indexes of all actions;

[0052] Step 603, using the customized physical test report generation model trained with historical physical test scores and language data, processing the current related data to generate the current physical test report text;

[0053] Step 604: The physical test report text includes the following contents: description of physical test results, description of movement analysis, and suggestions for movement improvement, which can help athletes improve their physical test results and movement skills;

[0054] Step 605, integrate the performance description, action analysis and improvement suggestions into the report template, polish the generated text, optimize the format of the report, and send it to the reviewer for verification.

[0055] In the second aspect, an automated score recording and standardized action evaluation system for campus sports project test analysis is used to implement a score recording and action evaluation method for campus sports project test analysis, including an end-side tablet computer, an end-side camera, a cloud-side server device, and an image processing algorithm, a deep learning model, a front-end reality interface, and a back-end database software. The end-side camera is communicatively connected to the end-side tablet computer, and the end-side tablet computer is communicatively connected to the cloud-side server. The end-side camera has a data acquisition module, and the end-side tablet device has a preliminary processing module and a key frame extraction module. The cloud-side server has an image processing module, a motion posture analysis module, a score calculation module, and a report generation module, wherein electrical signals are connected between each module;

[0056] The data acquisition module uses a depth camera and an RGB camera to collect video data in the campus physical test scene;

[0057] The terminal-side preliminary processing module is used to perform preliminary processing on the tablet device, obtain a human body depth map through human body depth estimation, perform 3D space modeling, perform preliminary calculation of physical test scores, and obtain information related to human body movement;

[0058] The key frame extraction module is used to determine the body measurement state according to the body depth information and extract the video frames of the key time period;

[0059] The cloud-side image processing module is used to process the RGB video on the cloud-side server by integrating the image processing algorithm and the deep learning model;

[0060] The score calculation module is used to calculate the physical test score and compare it with the preset standard to evaluate the quality of the score;

[0061] The action posture analysis module performs a detailed analysis of the action posture based on the results of the cloud-side image processing and evaluates the standardization of the action;

[0062] The report generation module is used to integrate the physical test results, movement posture evaluation results, and historical data into a physical test report using a customized report generation model.

[0063] Due to the adoption of the above technical solution, the present invention has the following technical advances compared with the prior art:

[0064] The present invention provides an automated score recording and standardized action evaluation method and system for testing and analyzing campus sports projects. The terminal-side equipment has low cost and strong reusability, and the cloud-side equipment can be flexibly expanded as needed. The evaluation results are fully automated and highly accurate, and can be displayed on a front-end tablet device in real time. The action evaluation dimensions are comprehensive, and the standardization and normativity of different actions in various physical test items can be evaluated. Combined with historical data, the results can be quickly returned to coaches and athletes. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] 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 recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0066] Figure 1 is a flow chart of the method of the present invention;

[0067] Figure 2 It is the overall system framework diagram of the present invention;

[0068] Figure 3 This is a flowchart of the image processing algorithm in the cloud-side server of the present invention;

[0069] Figure 4 A flowchart of obtaining information related to human body movement according to the present invention;

[0070] Figure 5 The present invention is a flow chart for calculating physical test results and evaluating movement posture. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0072] Embodiment 1, as Figure 1-5 As shown, the present invention provides an automated performance recording and standardized action evaluation method for campus sports project test analysis, comprising the following steps:

[0073] Step 1: Configure the end-side image acquisition and processing device and the cloud-side image processing device, and start the end-side device and the cloud-side device respectively, wherein the end-side device includes a depth camera, an RGB camera, and a tablet device, and the cloud-side device is a cloud-side server. Configure the end-side image acquisition and processing device, wherein the end-side image acquisition and processing device includes a depth camera, an RGB camera, and a tablet device. Place the end-side device with a tripod at a specific location next to the test area, adjust the camera's viewing angle and focal length so that the camera can capture images that cover the entire test area, and connect the depth camera to the tablet device via a wired connection. The RGB camera can be a built-in camera for the tablet device or an external device supporting the depth camera. Install the corresponding image acquisition and processing software on the tablet device, and configure the software. The device detects the presence of a camera and a tablet, and receives data from the depth camera and the RGB camera. The camera and the tablet are the end-side devices. The camera and the tablet are turned on to check the end-side devices. The cloud-side image processing device is the cloud-side server, and the cloud-side server is the cloud-side device. The operating system and software environment of the cloud-side server are installed and configured. The image processing and deep learning models are deployed on the cloud-side server, and the network connection between the cloud-side server and the end-side device is configured to receive data from the end-side device. The end-side device and the cloud-side device are turned on respectively to check whether the communication between the devices is normal. The tablet device and the camera are used in conjunction with the image acquisition and processing software to capture test data. The cloud-side server is used to receive the test data of the end-side device, and the data upload process is monitored.

[0074] Step 2, using a depth camera to collect campus physical test videos, obtain a human body depth map through human body depth estimation, and perform 3D space modeling based on the depth information, perform preliminary calculations on physical test scores, obtain information related to human body movements, use a depth camera to collect campus physical test videos, and analyze the depth image of the test area captured by the campus physical test video, perform preprocessing operations such as noise reduction and illumination compensation on the captured depth image of the test area, use the depth image data, detect the human body area from the depth image through an image processing algorithm, and extract the depth information of the human body area, based on the extracted depth information of the human body area, convert the key pixel points in the two-dimensional depth map into points in three-dimensional space to form point cloud data, and construct three-dimensional space modeling of human body key points through a surface reconstruction algorithm, according to the human body key point data in 3D space, combined with the rules of the physical test items, analyze the trajectory of human body movement and calculate preliminary scores;

[0075] Step 3: Display preliminary results on the client device for on-site personnel to view immediately.

[0076] Step 4: Determine the physical test status according to the human body depth information, determine the key time period, use the RGB camera to collect video clips of the key time period, and upload the preliminary score and RGB video clips generated by the end-side device to the cloud-side server, analyze the extracted human body area depth information, calculate the key movements and states of the human body, determine the physical test status of the person being tested, detect the key movements and states, determine the current moment as the beginning of the key time period, mark it, and determine the length of the key time period. Based on the determined key time period, use the RGB camera to collect video of the key time period to obtain the RGB video clip. On the end-side device, calculate the preliminary physical test score based on the data captured by the depth camera and the preset physical test rules, and package the preliminary physical test score and RGB video clips and upload them to the cloud-side server;

[0077] Step 5: In the cloud-side server, the uploaded video clips are processed using an image processing algorithm and a deep learning model based on RGB images, the physical test scores are calculated, and the action postures are evaluated. In the cloud-side server, the uploaded RGB video clips are decoded into image frames, and the image enhancement algorithm is applied to process the image frames decoded from the RGB video clips to improve the image quality. The human body detection model based on deep learning is used to detect the human body targets in each frame of the image, and the human body skeleton key point model based on deep learning is used to extract the human body's skeletal joints. The object detection model is used to detect the sports equipment in the image, and the equipment is segmented and classified to extract its key features. Combined with the image position information of the human body and the sports equipment, the human body skeleton key point information, and the actual size information of the human body and the sports equipment, it is determined whether the person being tested is in a valid state and a key action stage of the physical test. According to the rules of the physical test items, the physical test scores are calculated using the extracted information, and the angles of specific joints of the human body are calculated based on the human skeleton key points. The weighted action evaluation index is obtained by analysis to evaluate the accuracy and standardization of the action.

[0078] Furthermore, the expression of the single action evaluation index is:

[0079]

[0080]

[0081] in, For the The evaluation index of each action, For the The actual detected action of the action, For the The critical effective action of an action, For the The critical full score action of the action, is the action evaluation function, and the obtained value is the action evaluation value. and For , , Two bone vectors of three joints, where the joints can be any joints of the human body, including the head, neck, shoulder, elbow, hand, hip, knee, ankle, heel, toe, etc. Less than the effective threshold When the action evaluation value is Greater than the full score threshold When the evaluation index is 1, the action evaluation value Between the two thresholds, the evaluation index changes linearly between 0 and 1;

[0082] Furthermore, the calculation process of the weighted action evaluation index is:

[0083] According to the standard rules of physical test items, the action evaluation function is defined. By using this function, the action evaluation value can be obtained by calculating the angles of the joints of key parts of the human body.

[0084] From the video clips obtained by the RGB camera and the depth camera, computer vision technology is used to extract the location information of key points and analyze the standardization and regularity of the movements;

[0085] According to the action evaluation function and key point position information, the action evaluation value of each actual action is estimated, and a single action evaluation index is calculated. Finally, the weighted action evaluation index of all key actions is calculated through the weight matrix.

[0086] Combined with human position information and weighted action evaluation index of key actions, the trajectory and posture of human movement are analyzed to evaluate the standardization and normativeness of the overall action;

[0087] Furthermore, the expression of the weighted action evaluation index is:

[0088]

[0089]

[0090]

[0091] in, is the weighted action evaluation index, is the number of key actions, For the The weight of each action can be determined according to the importance of the action in different physical test items;

[0092] Step 6. Based on the data processed by the cloud-side server, a customized physical test report generation model is used to generate a physical test report and return it to the relevant personnel. All relevant data processed in the cloud-side server, including physical test scores and key action information, are summarized and converted to data standardization. The physical test scores are analyzed, and the key action information is compared with the preset standards to determine whether the action meets the standard requirements. The action standardization evaluation results are generated, including the action evaluation index of all actions. The customized physical test report generation model trained with historical physical test scores and language data is used to process the relevant data and generate the physical test report text. The physical test report text includes the following contents: physical test score description, action analysis description, and action improvement suggestions. These contents can help athletes improve their physical test scores and action skills. The score description, action analysis, and improvement suggestions are integrated into the report template, and the generated text is polished. At the same time, the format of the report is optimized and sent to the reviewer for verification.

[0093] Embodiment 2, as Figure 5 As shown, on the basis of Example 1, the present invention also provides an automated score recording and standardized action evaluation system for campus sports project test analysis, which is used to implement a score recording and action evaluation method for campus sports project test analysis, including devices such as a terminal-side tablet computer, a terminal-side camera, a cloud-side server, and image processing algorithms, deep learning models, front-end reality interfaces, back-end databases and other software, the terminal-side camera is communicatively connected to the terminal-side tablet computer, the terminal-side tablet computer is communicatively connected to the cloud-side server, the terminal-side camera has a data acquisition module, the terminal-side tablet device has a preliminary processing module, a key frame extraction module, the cloud-side server has an image processing module, a motion posture analysis module, a score calculation module and a report generation module, wherein the electrical signals between the modules are connected;

[0094] The data acquisition module uses a depth camera and an RGB camera to collect video data in the campus physical test scene;

[0095] The terminal-side preliminary processing module is used to perform preliminary processing on the tablet device, obtain a human depth map through human depth estimation, perform 3D space modeling, perform preliminary calculations on physical test scores, and obtain information related to human movement;

[0096] A key frame extraction module is used to determine the body measurement state according to the body depth information and extract the video frames in the key time period;

[0097] The cloud-side image processing module is used to process RGB videos on the cloud-side server by integrating image processing algorithms and deep learning models;

[0098] The score calculation module is used to calculate the physical test scores and compare them with the preset standards to evaluate the quality of the scores;

[0099] The action posture analysis module conducts detailed analysis of the action posture based on the results of cloud-side image processing and evaluates the standardization of the action;

[0100] The report generation module is used to integrate physical test scores, movement posture assessment results, and historical data into a physical test report using a customized report generation model.

[0101] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. The method for automated performance recording and standardized action evaluation for campus sports project test analysis is characterized by: The following steps are involved: Step 1: Configure the end-side image acquisition and processing device and the cloud-side image acquisition and processing device, and start the end-side device and the cloud-side device respectively. The end-side image acquisition and processing device includes a depth camera, an RGB camera, and a tablet device, and the cloud-side image acquisition and processing device is a cloud-side server. Step 2: Use a depth camera to collect campus physical test videos, obtain a human body depth map through human body depth estimation, perform 3D space modeling based on the depth information, perform preliminary calculations on the physical test scores, and obtain information related to human body movement; Step 3: Display preliminary results on the client device for on-site personnel to view immediately. Step 4: Determine the physical test status based on the human body depth information, determine the key time period, use the RGB camera to collect video clips in the key time period, and upload the preliminary scores and RGB video clips generated by the client device to the cloud server. Step 5: In the cloud-side server, the uploaded video clips are processed using an image processing algorithm based on RGB images and a deep learning model to calculate the physical test scores and evaluate the action postures. In step 5, the process of calculating the physical test scores and evaluating the action postures is as follows: Step 501: In the cloud-side server, the uploaded RGB video clip is decoded into image frames, and an image enhancement algorithm is applied to process the image frames decoded from the RGB video clip to improve image quality; Step 502, using a human body detection model based on deep learning to detect human body targets in each frame of image, and using a human body skeleton key point model based on deep learning to extract human body skeleton joint points; Step 503, using the object detection model to identify the sports equipment in the image, and segmenting and classifying the equipment to extract its key features; Step 504, combining the image position information of the human body and the sports equipment, the key point information of the human skeleton, and the actual size information of the human body and the sports equipment, to determine whether the person being measured is in a valid state and a key action stage of the physical test; Step 505, according to the rules of the physical test items, the physical test scores are calculated using the extracted information, and the angles of specific joints of the human body are calculated according to the key points of the human skeleton, and the weighted action evaluation index is obtained by analysis to evaluate the accuracy and standardization of the action; The calculation process of the weighted action evaluation index is: According to the standard rules of physical test items, the action evaluation function is defined. By using this function, the action evaluation value can be obtained by calculating the angles of the joints of key parts of the human body. From the video clips obtained by the RGB camera and the depth camera, computer vision technology is used to extract the location information of key points and analyze the standardization and regularity of the movements; According to the action evaluation function and key point position information, the action evaluation value of each actual action is estimated, and a single action evaluation index is calculated. Finally, the weighted action evaluation index of all key actions is calculated through the weight matrix. Combined with the weighted action evaluation index, the trajectory and posture of human movement are analyzed to evaluate the standardization and normativeness of the overall action; the expression of a single action evaluation index is: ； ； in, For the The evaluation index of each action, For the The actual detected action of the action, For the The critical effective action of an action, For the The critical full score action of the action, is the action evaluation function, and the obtained value is the action evaluation value. and Respectively are two vectors of the joint points; when the action evaluation value Less than the effective threshold When the action evaluation value is Greater than the full score threshold When the evaluation index is 1, the action evaluation value Between the two thresholds, the evaluation index changes linearly between 0 and 1; The expression of the weighted action evaluation index is: ； ； ； in, is the weighted action evaluation index, is the number of key points; Step 6: Based on the data processed by the cloud-side server, a customized physical test report generation model is used to generate a physical test report and return it to the relevant personnel.

2. The method for automated performance recording and standardized action evaluation of campus sports project test analysis according to claim 1 is characterized in that: In step 1, the process of configuring the terminal-side image acquisition and processing device and the cloud-side image acquisition and processing device is as follows: Step 101: configure the end-side image acquisition and processing device, wherein the end-side image acquisition and processing device includes a depth camera, an RGB camera, and a tablet device. The depth camera is installed in the test area to cover the entire test area. The viewing angle and focal length of the camera are adjusted to capture the required image. The depth camera is connected to the tablet device via a wired method. The RGB camera is a built-in camera of the tablet device, or an external device supporting the depth camera. Step 102: Install corresponding image acquisition and processing software on the tablet device, configure the software, and receive data from the depth camera and the RGB camera. The camera and the tablet device are the end-side devices. Turn on the camera and the tablet device to check the end-side devices. Step 103, configuring a cloud-side image acquisition and processing device, wherein the cloud-side image acquisition and processing device is a cloud-side server, and the cloud-side server is a cloud-side device, and installing and configuring an operating system and software environment of the cloud-side server; Step 104: deploy image processing and deep learning models on the cloud-side server, and configure a network connection between the cloud-side server and the end-side device to receive data from the end-side device; Step 105, respectively start the terminal side device and the cloud side device, check whether the communication between the devices is normal, use the tablet device and the camera in conjunction with the image acquisition and processing software to capture test data, use the cloud side server to receive the test data of the terminal side device, and monitor the data upload process.

3. The method for automated performance recording and standardized action evaluation of campus sports project test analysis according to claim 2 is characterized in that: In step 2, the process of obtaining the information related to human body movement is as follows: Step 201, using a depth camera to collect campus physical measurement videos, and analyzing the depth image of the test area captured by the campus physical measurement video, and performing pre-processing operations of noise reduction and illumination compensation on the captured depth image of the test area; Step 202, using the depth image data, detecting the human body region from the depth image through an image processing algorithm, and extracting the depth information of the human body region, including human skeleton key point data; Step 203, based on the extracted human body area depth information, convert the key pixel points in the two-dimensional depth map into points in three-dimensional space to form point cloud data, and construct a three-dimensional space model of the key points of the human body through a surface reconstruction algorithm; Step 204, based on the human body key point data in the 3D space and in combination with the rules of the physical test items, the trajectory of the human body movement is analyzed and the preliminary score is calculated.

4. The method for automated performance recording and standardized action evaluation of campus sports project test analysis according to claim 3 is characterized in that: In step 4, the process of collecting video clips in the key time period is as follows: Step 401, analyzing the extracted human body area depth information, detecting key actions and states of the human body area depth information, and determining the physical measurement state of the measured person; Step 402, detecting the key action and state, determining the current moment as the beginning of the key time period, marking it, and determining the length of the key time period; Step 403, based on the determined key time period, using an RGB camera to collect video of the key time period to obtain an RGB video clip; Step 404: On the terminal device, a preliminary physical test score is calculated based on the data captured by the depth camera and the preset physical test rules, and the preliminary physical test score and the RGB video clip are packaged and uploaded to the cloud server.

5. The method for automated performance recording and standardized action evaluation of campus sports project test analysis according to claim 4 is characterized in that: In step 6, the process of generating the physical test report is as follows: Step 601, summarizing all relevant data processed in the cloud-side server, including physical test scores and key action information, and performing data standardization conversion; Step 602, analyzing the physical test results, and comparing the key action information with the preset standards to determine whether the action meets the standard requirements, and generating action standardization evaluation results, including action evaluation indexes of all actions; Step 603, using the customized physical test report generation model trained with historical physical test scores and language data, processing the current related data to generate the current physical test report text; Step 604: The physical test report text includes the following contents: description of physical test results, description of movement analysis, and suggestions for movement improvement, which can help athletes improve their physical test results and movement skills; Step 605, integrate the performance description, action analysis and improvement suggestions into the report template, polish the generated text, optimize the format of the report, and send it to the reviewer for verification.

6. An automated score recording and standardized action evaluation system for campus sports project test analysis, used to implement the automated score recording and standardized action evaluation method for campus sports project test analysis described in any one of claims 1-5, comprising a terminal-side tablet computer, a terminal-side camera, a cloud-side server device, and an image processing algorithm, a deep learning model, a front-end reality interface, and a back-end database software, characterized in that: The end-side camera is communicatively connected to the end-side tablet computer, the end-side tablet computer is communicatively connected to the cloud-side server, the end-side camera has a data acquisition module, the end-side tablet computer has a preliminary processing module and a key frame extraction module, the cloud-side server has an image processing module, a motion posture analysis module, a score calculation module and a report generation module, wherein the modules are connected by electrical signals; The data acquisition module uses a depth camera and an RGB camera to collect video data in the campus physical test scene; The terminal-side preliminary processing module is used to perform preliminary processing on the tablet device, obtain a human body depth map through human body depth estimation, perform 3D space modeling, perform preliminary calculations on physical test scores, and obtain information related to human body movements; The key frame extraction module is used to determine the body measurement state according to the body depth information and extract the video frames of the key time period; The image processing module is used to process the RGB video on the cloud-side server by integrating the image processing algorithm and the deep learning model; The score calculation module is used to calculate the physical test score and compare it with the preset standard to evaluate whether the score meets the standard; The action posture analysis module performs a detailed analysis of the action posture based on the results of the cloud-side image processing and evaluates the standardization of the action; The report generation module is used to integrate the physical test results, movement posture evaluation results, and historical data into a physical test report using a customized report generation model.

Images