Campus sports digital management method and system

Abstract

The present invention discloses a campus sports digital management method and system thereof, which relates to the technical field of campus sports intelligent digital management. The method first collects students' basic sports information, target skill movement categories and teaching standard text data to establish a standard movement data set; collects video images, electromyographic signals and skin stretching data during the students' movement execution to construct a movement perception set; evaluates the movement quality by calculating the movement completion consistency index and comparing it with a first threshold; calculates the muscle coordination coefficient and compares it with a second threshold to monitor the movement force structure; based on the trained AI scoring model, automatically outputs the movement score and position error, further calculates the skill growth index and compares it with a third threshold to dynamically evaluate the student's skill growth trend; the system automatically generates personalized training strategies and comprehensive scoring reports based on the evaluation results, to achieve accurate monitoring and guidance of movement quality, force coordination and skill growth.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent digital management of campus sports, and in particular to a digital management method and system for campus sports. Background Art

[0002] With the rapid development of information technology and digital education, campus sports management is gradually incorporating digital tools to improve the efficiency of monitoring student athletic performance and teaching management. Existing digital campus sports management systems primarily rely on wearable devices or terminals to collect student performance data during physical activity, such as running distance, cadence, heart rate, and long jump scores. This data is suitable for basic physical fitness projects based primarily on time-space parameters and can, to a certain extent, assist teachers in assessing students' physical fitness.

[0003] However, traditional systems have significant shortcomings when it comes to complex, technically demanding sports like football, basketball, and gymnastics. Because these sports emphasize the quality of movement, timing logic, and coordination, single-source timing and measurement data cannot effectively reflect a student's actual performance during skill learning. For example, traditional systems cannot determine whether a student's shooting form is correct, whether gymnastics movements are executed correctly, or whether a pass meets technical standards.

[0004] Furthermore, existing methods generally lack the capabilities of semantic motion modeling and high-dimensional perceptual analysis, making them unable to accurately identify and evaluate complex sports skills, nor generate targeted feedback on skill growth. This limitation severely restricts the in-depth application of digital physical education in areas such as high-level skill training, individualized assessment, and long-term ability tracking. Summary of the Invention

[0005] In view of the deficiencies of the existing technology, the present invention provides a campus sports digital management method and system thereof to solve the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a campus sports digital management method and system thereof, comprising the following steps:

[0007] Step 1: Collect students' basic physical education information, target skill movement categories, and teaching standard text data. Based on the standard movement steps set by the teacher, build a movement semantic map, extract the key steps and timing logic of the skill movements, generate a movement sequence knowledge chain, and establish a standard movement dataset;

[0008] Step 2: Collect video image data, electromyographic signal data, and skin stretching data during the student's target action execution to construct an action perception set;

[0009] Step 3: Extract each frame of the continuous frame video image sequence to obtain the joint posture vector and the actual time point of the action node completion. Combined with the data of the standard action dataset, the action completion consistency index MCI is calculated and compared with the first threshold Q1 to determine whether the student's action quality meets the standard. If not, a strategy is given;

[0010] Step 4: By extracting the electromyographic signals and skin stretch of the action perception set and combining them with the data from the standard action dataset, the muscle coordination coefficient EMI is calculated and compared with the second threshold Q2 to determine whether the student's movement force structure is normal. If abnormal, a strategy is given;

[0011] Step 5. By establishing an AI scoring model, the trained model serves as a scoring engine to automatically output the student's action score Dsi and position error residual value Rvi; further calculate and obtain the skill growth index GRI, and compare and analyze it with the third threshold Q3 to determine whether the student's skill growth trend is qualified. If it is unqualified, a strategy is given, and a comprehensive student individual sports skill scoring report is generated.

[0012] Preferably, step one includes:

[0013] S11. Collect students' basic physical education information, including age, gender, grade, physical indicators, and the types of training target skills and movements, and establish student physical development files;

[0014] S12. Import the standard movement text data set by the teacher, including the structured content of the standard movement name, movement key points, execution requirements and scoring criteria;

[0015] S13. Based on the standard action text data, construct an action semantic graph and extract the key steps and timing logic of the skill action;

[0016] S14. Generate the corresponding action sequence knowledge chain based on the action semantic graph and establish a standard action dataset.

[0017] Preferably, step 2 includes:

[0018] S21. Install a high-definition motion camera to monitor the student's whole-body motion trajectory and key postures in real time during the execution of the target skill movements, and collect continuous frame video image sequences;

[0019] S22. Using a surface electromyography (SEM) device, monitor the electrophysiological activation process of the main muscle groups involved in the student's target movements in real time and collect electromyographic signals.

[0020] S23. Using a flexible skin strain sensor, the student's skin at the joint activity area is subjected to real-time monitoring of the stretching and compression changes and the skin stretchability is collected.

[0021] S24. Synchronously process the collected continuous frame video image sequence, electromyographic signals, and skin stretch, unify the time axis, remove noise, perform feature extraction, and establish a motion perception set.

[0022] Preferably, step three includes:

[0023] S31. Extract each frame of image from a continuous frame video image sequence, use the human posture recognition tool OpenPose to extract the student's joint position from each frame of image, and obtain the joint posture vector. Use the timestamp of the video frame to find the actual time point when each action node is completed. Combined with the data of the standard action dataset, after dimensionless processing, calculate the action completion consistency index (MCI).

[0024] Preferably, step three further includes:

[0025] S32, by presetting a first threshold Q1, and comparing and analyzing the action completion consistency index MCI with the first threshold Q1, obtaining a first evaluation result includes:

[0026] When the movement completion consistency index MCI ≥ the first threshold Q1, it means that the student's movement quality meets the standard and no adjustment is made, and continuous monitoring is performed;

[0027] When the movement completion consistency index MCI is less than the first threshold Q1, it means that the student's movement quality does not meet the standard and there is a deviation in the movement execution, which triggers the first warning instruction and generates the first strategy: provide targeted guidance and improvement to students, provide micro-movement correction prompts, mark the current movement stage as the key intervention section, and use AI to generate standard movement animations for comparative learning.

[0028] Preferably, step four includes:

[0029] S41. By extracting the electromyographic signals and skin stretchability of the motion perception set, combining them with the data of the standard motion data set, and performing dimensionless processing, the muscle coordination coefficient EMI is calculated.

[0030] Preferably, step four further includes:

[0031] S42, by presetting a second threshold value Q2 and comparing and analyzing the muscle coordination coefficient EMI with the second threshold value Q2, obtaining a second evaluation result includes:

[0032] When the motor coordination coefficient EMI ≥ the second threshold Q2, it indicates that the student's motor force structure is normal and no adjustments are made, but continuous monitoring is performed.

[0033] When the motor coordination coefficient EMI is less than the second threshold Q2, it indicates that the student's motor force structure is abnormal, and there is a risk of repetitive injury and abnormal force. The second warning instruction is triggered, and the second strategy is generated: the lesion fusion identification process is started for the student, the current training is suspended, the electromyography monitoring of the corresponding muscle group is refined and sampled, and the samples are submitted to the physical education teacher or rehabilitation instructor for manual review.

[0034] Preferably, step five includes:

[0035] S51. Construct an initial convolutional neural network model for action scoring by using a convolutional neural network, and use standard action data, the student's action perception set, and the corresponding action completion consistency index MCI and muscle coordination coefficient EMI as training features to train and test the initial convolutional neural network model; use the trained initial convolutional neural network model as an AI scoring model, extract the intermediate layer output in the model as a feature vector, and use it to identify the comprehensive feature information of action execution quality and muscle coordination; further train and test the AI ​​scoring model based on the obtained feature information to achieve a comprehensive scoring of the student's action quality; finally, use the trained AI scoring model as a scoring engine, input the student's action data of the current training cycle, the corresponding MCI and EMI, and the model automatically outputs: action score Dsi and position error residual value Rvi;

[0036] S52. By extracting the action score Dsi and position error residual value Rvi output by the AI ​​scoring model, combined with the comparison of the student's training cycle, and after dimensionless processing, the skill growth index GRI is calculated.

[0037] Preferably, step five further includes:

[0038] S53. By presetting a third threshold Q3 and comparing and analyzing the skill growth index GRI with the third threshold Q3, obtaining a third evaluation result includes:

[0039] When the skill growth index GRI ≥ the third threshold Q3, it means that the student's skill growth trend is qualified and no adjustment is made, and continuous monitoring is carried out;

[0040] When the skill growth index (GRI) is less than the third threshold Q3, it indicates that the student's skill growth trend is unsatisfactory, the training effect is not up to standard, and there is a learning bottleneck or training efficiency problem. This triggers the third warning instruction and generates the third strategy: based on the fluctuation of the student's position error residual value, training feedback reminders are generated; teachers and students are guided to focus on reviewing frequently incorrect movements, and auxiliary training content is automatically recommended: decomposition movement demonstration videos and key force teaching;

[0041] S54. Comprehensively generate individual sports skill scoring reports for students, including the growth of movement scores, skill growth curves, error location markings, and suggested correction paths, and automatically push them to the teacher and student ends. At the same time, all data are archived in the student's sports growth file.

[0042] Preferably, the campus sports digital management system includes:

[0043] The standard movement data set establishment module is used to collect students' basic physical education information, target skill movement categories, and teaching standard text data. Based on the standard movement steps set by the teacher, it constructs a movement semantic map, extracts the key steps and timing logic of the skill movements, generates a movement sequence knowledge chain, and establishes a standard movement data set;

[0044] The motion perception data acquisition module is used to collect video image data, electromyographic signal data, and skin stretching data during the student's target action execution to construct a motion perception set;

[0045] The action quality monitoring module is used to extract each frame of the continuous frame video image sequence, obtain the joint posture vector and the actual time point of the action node completion, and calculate the action completion consistency index MCI based on the data of the standard action dataset. It is then compared and analyzed with the first threshold Q1 to determine whether the student's action quality meets the standard. If not, a strategy is given;

[0046] The sports force monitoring module is used to extract the electromyographic signals and skin stretchability of the action perception set, combine them with the data of the standard action data set, calculate the muscle coordination coefficient EMI, and compare and analyze it with the second threshold Q2 to determine whether the student's sports force structure is normal. If it is abnormal, a strategy is given;

[0047] The AI ​​scoring model establishment and skill growth monitoring module is used to establish an AI scoring model. The trained model serves as a scoring engine to automatically output the student's movement score Dsi and position error residual value Rvi; further calculate and obtain the skill growth index GRI, and compare and analyze it with the third threshold Q3 to determine whether the student's skill growth trend is qualified. If it is unqualified, a strategy is given and a comprehensive student individual sports skill scoring report is generated.

[0048] The present invention provides a digital management method and system for campus sports, which has the following beneficial effects:

[0049] (1) The campus sports digital management method and system not only supports data collection and monitoring of conventional events such as running and long jump, but more importantly, by integrating video images, electromyographic signals and skin stretching data, it can accurately identify and quantify the completion quality and force structure of dynamic and complex skill movements such as football, basketball, and gymnastics, effectively filling the technical gap that traditional campus sports digital management methods cannot cover skill-related events.

[0050] (2) The campus sports digital management method and system, through the teaching standard data set by teachers, constructs an action semantic map with temporal logic and sub-task structure, generates an action sequence knowledge chain, and establishes a standard action data set, providing clear and structured knowledge reference for subsequent action recognition, deviation detection and scoring models, significantly improving recognition accuracy and model generalization ability.

[0051] (3) The digital management method and system of campus sports collects electromyographic signals and skin strain data to calculate the muscle coordination coefficient EMI. For the first time, it introduces the muscle coordination rhythm and movement tension state of students in skill movements into the evaluation system. It can not only judge the technical level, but also warn of potential force abnormalities or injury risks, and provide guidance strategies with more intervention value.

[0052] (4) The campus sports digital management method and system thereof, the present invention outputs action scores and error residuals through the AI ​​scoring model, and further calculates the skill growth index GRI, automatically judges the skill growth trend in combination with the set threshold, and finally outputs a personalized scoring report and growth curve chart, realizing intelligent evaluation and phased feedback closed loop of students' sports skill training, which helps teachers to provide precise guidance and students to make self-adjustments. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a schematic diagram of the steps of the campus sports digital management method of the present invention;

[0054] Figure 2 This is a block diagram and flow chart of the campus sports digital management system of the present invention. DETAILED DESCRIPTION

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 making creative efforts are within the scope of protection of the present invention.

[0056] Example 1

[0057] See also Figure 1The present invention provides a campus sports digital management method and system thereof, comprising the following steps:

[0058] Step 1: Collect students' basic physical education information, target skill movement categories, and teaching standard text data. Based on the standard movement steps set by the teacher, build a movement semantic map, extract the key steps and timing logic of the skill movements, generate a movement sequence knowledge chain, and establish a standard movement dataset;

[0059] Step 2: Collect video image data, electromyographic signal data, and skin stretching data during the student's target action execution to construct an action perception set;

[0060] Step 3: Extract each frame of the continuous frame video image sequence to obtain the joint posture vector and the actual time point of the action node completion. Combined with the data of the standard action dataset, the action completion consistency index MCI is calculated and compared with the first threshold Q1 to determine whether the student's action quality meets the standard. If not, a strategy is given;

[0061] Step 4: By extracting the electromyographic signals and skin stretch of the action perception set and combining them with the data from the standard action dataset, the muscle coordination coefficient EMI is calculated and compared with the second threshold Q2 to determine whether the student's movement force structure is normal. If abnormal, a strategy is given;

[0062] Step 5. By establishing an AI scoring model, the trained model serves as a scoring engine to automatically output the student's action score Dsi and position error residual value Rvi; further calculate and obtain the skill growth index GRI, and compare and analyze it with the third threshold Q3 to determine whether the student's skill growth trend is qualified. If it is unqualified, a strategy is given, and a comprehensive student individual sports skill scoring report is generated.

[0063] In this embodiment, by introducing the movement completion consistency index MCI and the muscle coordination coefficient EMI, an accurate quantitative analysis of the quality of students' sports skill execution is achieved. It can not only evaluate whether the movements are standard, but also identify potential sports risks such as force rhythm and joint tension. Therefore, personalized intervention strategies are automatically generated when students' movements do not meet the standards or the force is abnormal, which significantly improves the scientific nature and pertinence of teaching feedback and effectively reduces the risk of training injuries caused by improper posture or uncoordinated force.

[0064] Example 2

[0065] This embodiment is explained in Example 1. Specifically, step 1 includes:

[0066] S11. Collect students' basic physical education information, including age, gender, grade, physical indicators, and the types of training target skills and movements, and establish student physical development files;

[0067] S12. Import the standard movement text data set by the teacher, including the structured content of the standard movement name, movement key points, execution requirements and scoring criteria;

[0068] S13. Based on the standard action text data, construct an action semantic graph and extract the key steps and timing logic of the skill action;

[0069] S14. Generate the corresponding action sequence knowledge chain based on the action semantic graph and establish a standard action dataset.

[0070] Collection of students' basic information, including age, gender, grade, physical indicators such as height, weight, flexibility assessment results, etc., as well as the category of current training target skill movements.

[0071] Import teaching standard data, import the skill movement teaching standard text data set by the teacher, including standard movement name, movement key points, execution requirements, scoring standards and other structured content.

[0072] The action semantic database is constructed based on the collected skill names, action instructions and standard descriptions to build an action semantic database in a unified format, converting the original text into structured semantic units such as "preparation posture", "force phase", "finishing action", etc.

[0073] Standard action graphs are generated based on the standard action steps set by the teacher and the timing logic of the skill actions to construct an action semantic graph, clarifying the key action nodes, sequence constraints and stage-by-stage action goals of each skill.

[0074] Skill action extraction: extract the action subtask structure corresponding to each skill from the action semantic map. For example, the "throwing" skill can be decomposed into subtask units such as "pull-up → backswing → release → follow-up".

[0075] The action timing chain is constructed to generate an action sequence knowledge chain based on the execution order of subtasks, time duration relationship, key frame logic, etc., and a standard action dataset is established to guide the subsequent model's timing recognition and action anchoring.

[0076] In this embodiment, by constructing an action semantic graph and extracting the subtask structure and timing chain of standard actions, the structured expression and knowledge modeling of sports skill actions are realized, so that complex skill actions are decomposed into subtask units with logical sequences and clear key frames, effectively supporting the subsequent action recognition and scoring models to accurately anchor and dynamically analyze the action execution process, and significantly improving the clarity of skill demonstration and the pertinence of training guidance in physical education teaching.

[0077] Example 3

[0078] This embodiment is explained in Example 2. Specifically, step 2 includes:

[0079] S21. Install a high-definition motion camera to monitor the student's whole-body motion trajectory and key postures in real time during the execution of the target skill movements, and collect continuous frame video image sequences for subsequent skeleton key point extraction and posture recognition;

[0080] S22. Use a surface electromyography (SEM) device to monitor the electrophysiological activation of the main muscle groups involved in the student's target movements in real time, collect EMG signals, and reflect changes in muscle force timing and intensity.

[0081] S23. Using flexible skin strain sensors, real-time monitoring of the stretching and compression changes in the skin of the student's joint activity area is performed to collect skin stretchability, which is used to assist in determining the joint motion range and dynamic tension state.

[0082] S24. Synchronously process the collected continuous frame video image sequence, electromyographic signals, and skin stretch, unify the time axis, remove noise, perform feature extraction, and establish a motion perception set.

[0083] In this embodiment, high-definition motion cameras, surface electromyography collectors and flexible skin strain sensors are deployed in a coordinated manner to collect multimodal data such as visual images, muscle force and joint dynamic tension during the students' target skill movements. Through synchronous processing and feature extraction on a unified time axis, a motion perception set integrating structured information is constructed, which significantly improves the comprehensive perception ability and dynamic precision analysis level of the students' actual motion execution status, providing a highly reliable data foundation for subsequent motion quality assessment and error diagnosis.

[0084] Example 4

[0085] This embodiment is explained in Example 3. Specifically, step 3 includes:

[0086] S31. Extract each frame from a continuous frame video image sequence. Use the human posture recognition tool OpenPose to extract the student's joint position from each frame to obtain the joint posture vector. Use the timestamp of the video frame to find the actual time point when each action node is completed. Combined with the data of the standard action dataset, after dimensionless processing, calculate the action completion consistency index (MCI). The formula is as follows:

[0087] ;

[0088] Where N represents the number of key action nodes in the standard semantic graph, and each node represents a complete standard action subtask. represents the human joint posture vector of the i-th action node performed by the student, represents the reference action posture vector in the standard semantic graph corresponding to the i-th action node, represents the actual time point when the student completes the action corresponding to the i-th semantic action node, represents the theoretical completion time point of the i-th standard action node in the standard graph, w1 represents the weight of the spatial deviation term, and w2 represents the weight of the time synchronization term.

[0089] The method for obtaining w1 and w2 is based on experimental statistical analysis of a large amount of student movement execution data and the Movement Completion Consistency Index (MCI). Regression fitting and sensitivity analysis of the impact weights of spatial deviation and temporal synchronization were performed using multiple sets of movement samples. The appropriate weight distribution ratio was determined by combining the experience and judgment of sports movement recognition experts and data scientists. Reference was made to movement quality evaluation standards and relevant movement recognition model tuning specifications, which generally guide the relative importance of spatial error and temporal deviation in the overall movement score. This weighting is used to scientifically balance spatial posture accuracy and movement temporal coordination, improve the comprehensive evaluation accuracy and application effectiveness of the Movement Consistency Index (MCI), and ensure the rationality and effectiveness of movement quality judgment.

[0090] In this embodiment, by processing continuous frame video images, using posture recognition tools such as OpenPose to extract the student's joint posture vectors, combining the time point of movement completion with the standard movement graph for comparison, the movement completion consistency index MCI is calculated, which can simultaneously quantify the degree of deviation of the student's movement space posture and time rhythm, effectively improving the fine-grained quantitative evaluation ability of the quality of students' skill movement execution, and providing a highly interpretable and clear feedback evaluation basis for accurately determining whether their movements meet the standards.

[0091] Example 5

[0092] This embodiment is explained in Example 4. Specifically, step 3 further includes:

[0093] S32, by presetting a first threshold Q1, and comparing and analyzing the action completion consistency index MCI with the first threshold Q1, obtaining a first evaluation result includes:

[0094] When the movement completion consistency index MCI ≥ the first threshold Q1, it means that the student's movement quality meets the standard and no adjustment is made, and continuous monitoring is performed;

[0095] When the movement completion consistency index MCI is less than the first threshold Q1, it means that the student's movement quality does not meet the standard and there is a deviation in the movement execution, which triggers the first warning instruction and generates the first strategy: provide targeted guidance and improvement to students, provide micro-movement correction prompts, mark the current movement stage as the key intervention section, and use AI to generate standard movement animations for comparative learning.

[0096] The first threshold, Q1, is obtained by statistically analyzing consistency data from a large number of student movements, extracting the distribution range of the Movement Completion Consistency Index (MCI) between movements that meet the standard and those that do not. This is then combined with the experience and judgment of physical education experts and movement analysts to determine a reasonable first threshold. This is based on reference to relevant sports training standards and movement quality evaluation specifications, which typically provide threshold ranges for movement consistency. This threshold is used to effectively distinguish whether a student's movement quality meets the standard, providing precise guidance for movement correction and training optimization.

[0097] In this embodiment, by setting the first threshold Q1 and comparing the movement completion consistency index MCI with it, an automatic judgment mechanism for the student's movement quality is realized; when the movement quality does not meet the standard, the system can immediately trigger the first warning instruction and generate a personalized improvement strategy including micro-movement correction prompts and AI standard movement animation comparative learning, thereby effectively improving students' cognitive ability and correction efficiency of their own movement deviations, and strengthening the adaptive feedback and refined intervention effects in physical skills teaching.

[0098] Example 6

[0099] This embodiment is explained in Example 5. Specifically, step 4 includes:

[0100] S41. By extracting the electromyographic signals and skin stretch of the motion perception set, combining them with the data of the standard motion dataset, and performing dimensionless processing, the muscle coordination coefficient EMI is calculated. The formula is as follows:

[0101] ;

[0102] In the formula, M represents the number of key muscle groups that students use to exert force. It represents the electrical timing signal of the jth muscle group when the student exerts force, represents the reference standard muscle group electrical timing signal corresponding to the j-th muscle group, It represents the skin stretching degree corresponding to the jth muscle group when the student exerts force, represents the reference standard skin stretch corresponding to the j-th muscle group, represents the Pearson correlation coefficient between the muscle group electrical signal collected by the student at the jth muscle group and the reference muscle group electrical signal of the corresponding muscle group in the standard action sample, which is used to evaluate the consistency of the muscle force rhythm. represents the Pearson correlation coefficient between the student's skin stretch corresponding to the j-th muscle group and the reference skin stretch in the standard sample, which is used to evaluate the consistency between the student's local movement tension and posture drive. a1 and a2 represent weight coefficients.

[0103] How a1 and a2 are obtained: Based on experimental statistical analysis of a large amount of student EMG and skin stretch data, the correlation distribution characteristics of EMG and skin stretch signals under different movement skills are extracted. Combined with the experience and judgment of sports physiology experts and motion analysts, a reasonable weight ratio of EMG and skin stretch coordination is determined. Reference is made to the sports rehabilitation and movement coordination evaluation standards, as well as the technical specifications for multimodal biosignal fusion. These standards generally provide guidance on how to scientifically balance the contributions of EMG activation and skin tension signals in comprehensive coordination evaluation. This weight setting is used to effectively integrate EMG and skin stretch data, improve the accuracy and sensitivity of the muscle coordination coefficient EMI, and thus more accurately evaluate the coordination and health status of students' movement force structure.

[0104] In this embodiment, by constructing the muscle coordination coefficient EMI, integrating and analyzing the electromyographic activation rhythm and skin stretching characteristics of the student's target muscle group, and introducing the Pearson correlation coefficient with the standard sample for similarity evaluation, the quantitative consistency judgment of the student's muscle force structure and joint tension state is achieved; this method effectively enhances the sports movement evaluation system's perception of force coordination and movement driving logic, and provides a more physiologically based analysis basis for subsequent intelligent feedback and precise teaching.

[0105] Example 7

[0106] This embodiment is explained in Example 6. Specifically, step 4 also includes:

[0107] S42, by presetting a second threshold value Q2 and comparing and analyzing the muscle coordination coefficient EMI with the second threshold value Q2, obtaining a second evaluation result includes:

[0108] When the motor coordination coefficient EMI ≥ the second threshold Q2, it indicates that the student's motor force structure is normal and no adjustments are made, but continuous monitoring is performed.

[0109] When the motor coordination coefficient EMI is less than the second threshold Q2, it indicates that the student's motor force structure is abnormal, and there is a risk of repetitive injury and abnormal force. There is a significant misalignment between the electromyographic signal and the skin stretch signal, and the force of different muscle groups is disordered. There may be hidden sports injuries, fatigue accumulation or technical pathology. The second early warning instruction is triggered and the second strategy is generated: the lesion fusion identification process is started for the student, the current training is suspended, the electromyographic monitoring of the corresponding muscle groups is refined and sampled, and the samples are submitted to the physical education teacher or rehabilitation instructor for manual review.

[0110] The second threshold, Q2, is determined by statistically analyzing a large number of student electromyographic signals and skin stretch data to extract the distribution characteristics of the motor coordination coefficient (EMI) under normal and abnormal force conditions. This is combined with the expertise of sports rehabilitation experts and biomechanics researchers to determine a reasonable second threshold. This is based on reference to relevant sports medicine standards and electromyographic analysis technical specifications, which typically provide threshold ranges for determining motor coordination. This threshold is used to effectively distinguish between normal and abnormal force structures in students' movements and to prevent the risk of sports injuries.

[0111] In this embodiment, by setting the second threshold Q2 and comparing and analyzing the muscle coordination coefficient EMI with it, it is possible to automatically identify abnormal states of the student's sports force structure, and trigger the lesion warning mechanism in real time when muscle force dislocation or tension imbalance occurs, effectively preventing hidden sports injuries and fatigue accumulation caused by technical movement disorders, achieving early intervention and precise protection in the training process, and ensuring students' sports health and safe development of skills.

[0112] Example 8

[0113] This embodiment is explained in Example 7. Specifically, step 5 includes:

[0114] S51. Construct an initial convolutional neural network model for action scoring by using a convolutional neural network, and use standard action data, student action perception set, and the corresponding action completion consistency index MCI and muscle coordination coefficient EMI as training features to train and test the initial convolutional neural network model; use the trained initial convolutional neural network model as an AI scoring model, extract the middle layer output in the model as a feature vector, and use it to identify the comprehensive feature information of action execution quality and muscle coordination; further train and test the AI ​​scoring model based on the obtained feature information to achieve a comprehensive score for the student's action quality; and finally use the trained AI The scoring model, serving as a scoring engine, takes as input the student's movement data from the current training cycle, along with the corresponding MCI and EMI. The model automatically outputs: a movement score Dsi, which characterizes the overall quality of movement completion; and a residual error value Rvi, which characterizes the degree of deviation from the standard movement in key body parts. This scoring model not only relies on traditional perception data such as visible light images and skeletal key point sequences, but also integrates semantic consistency information (provided by MCI) and physiological force coordination information (provided by EMI). During the model training phase, structural indicators are used to guide the model's perception of "correct movement" and "whether the body is exerting the right force," thereby improving scoring accuracy and targeted feedback.

[0115] S52. By extracting the action score Dsi and the position error residual value Rvi output by the AI ​​scoring model, combined with the student's training cycle comparison, and dimensionless processing, the skill growth index GRI is calculated. The formula is as follows:

[0116] ;

[0117] Where T represents the total number of cycles of movement training that students participate in, represents the growth of students’ action scores in period t, ; represents the change in the student's position error in the tth period, ; and Represents the weight coefficient.

[0118] and How to obtain it: Through statistical analysis of the changes in movement scores and position error fluctuations in a large amount of student sports training data, the weight distribution of the influence of score growth and error changes in different training stages is extracted. Combined with the experience and judgment of sports teaching experts and sports science researchers, a reasonable weight coefficient ratio is determined. Referring to the sports skill evaluation standards and training effect feedback mechanism, these standards guide the setting of weights to scientifically reflect the relative importance of movement quality improvement and error control. This weight coefficient is used to reasonably adjust the contribution ratio of score growth and error reduction in the skill growth index (GRI), thereby more accurately evaluating the comprehensive growth trend of students' movement skills and assisting in the development of personalized training plans.

[0119] In this embodiment, a multimodal convolutional neural network scoring model is constructed that integrates the movement consistency index MCI and the muscle coordination coefficient EMI to achieve a comprehensive evaluation of students' movement quality and muscle coordination. It can accurately identify subtle deviations and force abnormalities in movement execution, and dynamically calculate the skill growth index GRI based on the training cycle, effectively reflecting the trend of students' skill improvement, thereby providing a scientific quantitative basis for personalized teaching, significantly improving the accuracy of scoring and the pertinence of feedback, and helping students to continuously optimize their sports skill performance.

[0120] Example 9

[0121] This embodiment is explained in Example 8. Specifically, step five further includes:

[0122] S53. By presetting a third threshold Q3 and comparing and analyzing the skill growth index GRI with the third threshold Q3, obtaining a third evaluation result includes:

[0123] When the skill growth index GRI ≥ the third threshold Q3, it means that the student's skill growth trend is qualified and the quality of the student's movement performance is gradually improving. No adjustment is made and continuous monitoring is carried out;

[0124] When the skill growth index (GRI) is less than the third threshold Q3, it indicates that the student's skill growth trend is unsatisfactory, the training effect is not up to standard, and there is a learning bottleneck or training efficiency problem. This triggers the third warning instruction and generates the third strategy: based on the fluctuation of the student's position error residual value, training feedback reminders are generated; teachers and students are guided to focus on reviewing frequently incorrect movements, and auxiliary training content is automatically recommended: decomposition movement demonstration videos and key force teaching;

[0125] The third threshold, Q3, is determined by statistically analyzing historical data on the Skill Growth Index (GRI) across a large number of student training cycles, extracting the distribution range of the index for qualified and unqualified skill growth trends. This is then combined with the experience and judgment of physical education experts and data analysts to determine a reasonable third threshold. This is based on reference to physical ability evaluation systems and growth monitoring standards, which typically provide a passing range for skill growth trends. This threshold is used to scientifically assess student skill growth trends and assist in optimizing training plans and personalized coaching programs.

[0126] S54. Comprehensively generate individual sports skill scoring reports for students, including the growth of movement scores, skill growth curves, error location markings, and suggested correction paths, and automatically push them to the teacher and student ends. At the same time, all data are archived in the student's sports growth file.

[0127] In this embodiment, by setting the skill growth index threshold Q3, dynamic monitoring and intelligent evaluation of students' skill growth trends can be achieved, training bottlenecks and inefficiencies can be discovered in a timely manner, personalized early warnings and precise tutoring strategies can be triggered, targeted training resources and improvement suggestions can be automatically pushed, and detailed skill scoring reports can be generated to assist teachers in scientific guidance and students in efficient improvement, significantly improving the sustainability of training effects and the level of individualized management.

[0128] Example 10

[0129] Campus Sports Digital Management System, please refer to Figure 2 ,include:

[0130] The standard movement data set establishment module is used to collect students' basic physical education information, target skill movement categories, and teaching standard text data. Based on the standard movement steps set by the teacher, it constructs a movement semantic map, extracts the key steps and timing logic of the skill movements, generates a movement sequence knowledge chain, and establishes a standard movement data set;

[0131] The motion perception data acquisition module is used to collect video image data, electromyographic signal data, and skin stretching data during the student's target action execution to construct a motion perception set;

[0132] The action quality monitoring module is used to extract each frame of the continuous frame video image sequence, obtain the joint posture vector and the actual time point of the action node completion, and calculate the action completion consistency index MCI based on the data of the standard action dataset. It is then compared and analyzed with the first threshold Q1 to determine whether the student's action quality meets the standard. If not, a strategy is given;

[0133] The sports force monitoring module is used to extract the electromyographic signals and skin stretchability of the action perception set, combine them with the data of the standard action data set, calculate the muscle coordination coefficient EMI, and compare and analyze it with the second threshold Q2 to determine whether the student's sports force structure is normal. If it is abnormal, a strategy is given;

[0134] The AI ​​scoring model establishment and skill growth monitoring module is used to establish an AI scoring model. The trained model serves as a scoring engine to automatically output the student's movement score Dsi and position error residual value Rvi; further calculate and obtain the skill growth index GRI, and compare and analyze it with the third threshold Q3 to determine whether the student's skill growth trend is qualified. If it is unqualified, a strategy is given and a comprehensive student individual sports skill scoring report is generated.

[0135] In this embodiment, the modular design of this system realizes a full-process closed loop from standard movement data construction, movement perception collection, real-time monitoring of movement quality and exercise force, to comprehensive skill growth evaluation based on AI scoring model. It can accurately capture students' movement execution details and muscle synergy, dynamically feedback movement quality and growth trends, provide personalized training strategies and scientific guidance, and significantly improve the efficiency and effectiveness of sports skills training.

[0136] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by technicians in this field for each set of sample data; as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0137] The above formulas are obtained by collecting a large amount of data and performing software simulation, and a formula close to the actual value is selected. The coefficients in the formula are set by those skilled in the art according to actual conditions. The above is only a preferred specific implementation method of the present invention, but the protection scope of the present invention is not limited to this. Any technician familiar with this technical field, within the technical scope disclosed by the present invention, can make equivalent replacements or changes based on the technical solution and inventive concept of the present invention, which should be covered by the protection scope of the present invention.

Claims

1. The digital management method of campus sports is characterized by: The following steps are involved: Step 1: Collect students' basic physical education information, target skill movement categories, and teaching standard text data. Based on the standard movement steps set by the teacher, build a movement semantic map, extract the key steps and timing logic of the skill movements, generate a movement sequence knowledge chain, and establish a standard movement dataset; Step 2: Collect video image data, electromyographic signal data, and skin stretching data during the student's target action execution to construct an action perception set; Step 3: Extract each frame of the continuous frame video image sequence to obtain the joint posture vector and the actual time point of the action node completion. Combined with the data of the standard action dataset, the action completion consistency index MCI is calculated and compared with the first threshold Q1 to determine whether the student's action quality meets the standard. If not, a strategy is given; The movement consistency index MCI is as follows: Where N represents the number of key action nodes in the standard semantic graph, and each node represents a complete standard action subtask. represents the human joint posture vector of the i-th action node performed by the student, represents the reference action posture vector in the standard semantic graph corresponding to the i-th action node, represents the actual time point when the student completes the action corresponding to the i-th semantic action node, represents the theoretical completion time point of the i-th standard action node in the standard graph, w1 represents the weight of the spatial deviation term, and w2 represents the weight of the time synchronization term; Step 4: By extracting the electromyographic signals and skin stretch of the action perception set and combining them with the data from the standard action dataset, the muscle coordination coefficient EMI is calculated and compared with the second threshold Q2 to determine whether the student's movement force structure is normal. If abnormal, a strategy is given; The EMI coefficient is as follows: In the formula, M represents the number of key muscle groups that students use to exert force. It represents the electrical timing signal of the jth muscle group when the student exerts force, represents the reference standard muscle group electrical timing signal corresponding to the j-th muscle group, It represents the skin stretching degree corresponding to the jth muscle group when the student exerts force, represents the reference standard skin stretch corresponding to the j-th muscle group, represents the Pearson correlation coefficient between the muscle group electrical signal collected by the student at the jth muscle group and the reference muscle group electrical signal of the corresponding muscle group in the standard action sample, which is used to evaluate the consistency of the muscle force rhythm. represents the Pearson correlation coefficient between the student's skin stretch corresponding to the jth muscle group and the reference skin stretch in the standard sample, which is used to evaluate the consistency between the student's local movement tension and posture drive. a1 and a2 represent weight coefficients; Step 5: By establishing an AI scoring model, the trained model serves as a scoring engine to automatically output the student's action score Dsi and position error residual value Rvi; further calculate and obtain the skill growth index GRI, and compare and analyze it with the third threshold Q3 to determine whether the student's skill growth trend is qualified. If it is unqualified, a strategy is given, and a comprehensive individual sports skill score report is generated for the student; The residual value of the part error Rvi is used to characterize the degree of deviation from the standard action on key body parts; The Skill Growth Index (GRI) is formulated as follows: Where T represents the total number of cycles of movement training that students participate in, represents the growth of students’ action scores in period t, represents the change in the student's position error in the tth period, and Represents the weight coefficient.

2. The campus sports digital management method according to claim 1, characterized in that: Step one includes: S11. Collect students' basic physical education information, including age, gender, grade, physical indicators, and the types of training target skills and movements, and establish student physical development files; S12. Import the standard movement text data set by the teacher, including the structured content of the standard movement name, movement key points, execution requirements and scoring criteria; S13. Based on the standard action text data, construct an action semantic graph and extract the key steps and timing logic of the skill action; S14. Generate the corresponding action sequence knowledge chain based on the action semantic graph and establish a standard action dataset.

3. The campus sports digital management method according to claim 2, characterized in that: Step 2 includes: S21. Install a high-definition motion camera to monitor the student's whole-body motion trajectory and key postures in real time during the execution of the target skill movements, and collect continuous frame video image sequences; S22. Using a surface electromyography (SEM) device, monitor the electrophysiological activation process of the muscle groups involved in the student's target movements in real time and collect electromyographic signals. S23. Using a flexible skin strain sensor, the student's skin at the joint activity area is subjected to real-time monitoring of the stretching and compression changes and the skin stretchability is collected. S24. Synchronously process the collected continuous frame video image sequence, electromyographic signals, and skin stretch, unify the time axis, remove noise, perform feature extraction, and establish a motion perception set.

4. The campus sports digital management method according to claim 3 is characterized in that: Step three includes: S31. Extract each frame of image from a continuous frame video image sequence, use the human posture recognition tool OpenPose to extract the student's joint position from each frame of image, and obtain the joint posture vector. Use the timestamp of the video frame to find the actual time point when each action node is completed. Combined with the data of the standard action dataset, after dimensionless processing, calculate the action completion consistency index (MCI).

5. The campus sports digital management method according to claim 4 is characterized in that: Step three also includes: S32, by presetting a first threshold Q1, and comparing and analyzing the action completion consistency index MCI with the first threshold Q1, obtaining a first evaluation result includes: When the movement completion consistency index MCI ≥ the first threshold Q1, it means that the student's movement quality meets the standard and no adjustment is made, and continuous monitoring is performed; When the movement completion consistency index MCI is less than the first threshold Q1, it means that the student's movement quality does not meet the standard and there is a deviation in the movement execution, which triggers the first warning instruction and generates the first strategy: provide targeted guidance and improvement to students, provide micro-movement correction prompts, mark the current movement stage as the key intervention section, and use AI to generate standard movement animations for comparative learning.

6. The campus sports digital management method according to claim 1, characterized in that: Step 4 also includes: By presetting the second threshold value Q2 and comparing and analyzing the muscle coordination coefficient EMI with the second threshold value Q2, obtaining the second evaluation result includes: When the motor coordination coefficient EMI ≥ the second threshold Q2, it indicates that the student's motor force structure is normal and no adjustments are made, but continuous monitoring is performed. When the motor coordination coefficient EMI is less than the second threshold Q2, it indicates that the student's motor force structure is abnormal, and there is a risk of repetitive injury and abnormal force. The second warning instruction is triggered, and the second strategy is generated: the lesion fusion identification process is started for the student, the current training is suspended, the electromyography monitoring of the corresponding muscle group is refined and sampled, and the samples are submitted to the physical education teacher or rehabilitation instructor for manual review.

7. The campus sports digital management method according to claim 6, characterized in that: Step five includes: S51. Construct an initial convolutional neural network model for action scoring by using a convolutional neural network, and use standard action data, student action perception set, and corresponding action completion consistency index MCI and muscle coordination coefficient EMI as training features to train and test the initial convolutional neural network model; use the trained initial convolutional neural network model as an AI scoring model, extract the middle layer output in the model as a feature vector, and use it to identify the comprehensive feature information of action execution quality and muscle coordination; further train and test the AI ​​scoring model based on the acquired feature information to achieve a comprehensive scoring of the student's action quality; finally, use the trained AI scoring model as a scoring engine, input the action data of the student's current training cycle, the corresponding MCI and EMI, and the model automatically outputs: action score Dsi and position error residual value Rvi; S52. Extract the action score Dsi and position error residual value Rvi output by the AI ​​scoring model, combine them with the student's training cycle comparison, and perform dimensionless processing to calculate the skill growth index GRI.

8. The campus sports digital management method according to claim 7, characterized in that: Step five also includes: S53. By presetting a third threshold Q3 and comparing and analyzing the skill growth index GRI with the third threshold Q3, obtaining a third evaluation result includes: When the skill growth index GRI ≥ the third threshold Q3, it means that the student's skill growth trend is qualified and no adjustment is made, and continuous monitoring is carried out; When the skill growth index (GRI) is less than the third threshold Q3, it indicates that the student's skill growth trend is unsatisfactory, the training effect is not up to standard, and there is a learning bottleneck or training efficiency problem. This triggers the third warning instruction and generates the third strategy: based on the fluctuation of the student's position error residual value, training feedback reminders are generated; teachers and students are guided to focus on reviewing frequently incorrect movements, and auxiliary training content is automatically recommended: decomposition movement demonstration videos and key force teaching; S54. Comprehensively generate individual sports skill scoring reports for students, including the growth of movement scores, skill growth curves, error location markings, and suggested correction paths, and automatically push them to the teacher and student ends. At the same time, all data are archived in the student's sports growth file.

9. Campus sports digital management system, according to the campus sports digital management method according to any one of claims 1 to 8, characterized in that: include: The standard movement data set establishment module is used to collect students' basic physical education information, target skill movement categories, and teaching standard text data. Based on the standard movement steps set by the teacher, it constructs a movement semantic map, extracts the key steps and timing logic of the skill movements, generates a movement sequence knowledge chain, and establishes a standard movement data set; The motion perception data acquisition module is used to collect video image data, electromyographic signal data, and skin stretching data during the student's target action execution to construct a motion perception set; The action quality monitoring module is used to extract each frame of the continuous frame video image sequence, obtain the joint posture vector and the actual time point of the action node completion, and calculate the action completion consistency index MCI based on the data of the standard action dataset. It is then compared and analyzed with the first threshold Q1 to determine whether the student's action quality meets the standard. If not, a strategy is given; The sports force monitoring module is used to extract the electromyographic signals and skin stretchability of the action perception set, combine them with the data of the standard action data set, calculate the muscle coordination coefficient EMI, and compare and analyze it with the second threshold Q2 to determine whether the student's sports force structure is normal. If it is abnormal, a strategy is given; The AI ​​scoring model establishment and skill growth monitoring module is used to establish an AI scoring model. The trained model serves as a scoring engine to automatically output the student's movement score Dsi and position error residual value Rvi; further calculate and obtain the skill growth index GRI, and compare and analyze it with the third threshold Q3 to determine whether the student's skill growth trend is qualified. If it is unqualified, a strategy is given and a comprehensive student individual sports skill scoring report is generated.

Images