Intelligent rope skipping sports safety warning method and computer equipment based on behavior recognition

Through the intelligent rope skipping movement safety warning method based on behavior recognition, using video data analysis and data traceability relationship network, the shortcomings of rope skipping safety monitoring in the existing technology are solved, real-time safety monitoring and risk warning of rope skipping activities are realized, and the sports are improved Security.

CN118823879BActive Publication Date: 2024-11-19ZHUHAI XINWEI TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411274833.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-12
Publication Date
2024-11-19
Estimated Expiration
2044-09-12

AI Technical Summary

Technical Problem

The existing safety monitoring methods for skipping ropes are difficult to monitor the skipping ropes in real time and accurately. Especially in large-scale physical education classes or extracurricular activities, it is difficult for teachers to pay attention to the safety of skipping ropes in each student at the same time, and the existing sports analysis system cannot be effective. Identify and alert potential security risks.

Method used

Using the intelligent rope skipping movement safety warning method based on behavior recognition, by collecting video data, extracting key features, building a data traceability relationship network, designing data inspection mechanism and triggering mechanism, dynamically adjusting risk assessment parameters, calculating comprehensive safety indicators, and performing early warnings and Suggested measures.

Benefits of technology

It realizes safety monitoring of rope skipping activities, effectively identify and warn of potential safety risks, provides personalized safety suggestions, and improves the safety of rope skipping.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118823879B_ABST
    Figure CN118823879B_ABST
Patent Text Reader

Abstract

The present application relates to an intelligent rope skipping sports safety warning method and computer equipment based on behavior recognition. The method includes: collecting video data of rope jumpers during rope skipping activities and preprocessing them; extracting key features from the preprocessed video data, and obtaining rope skipping status categories based on the key features; building a data traceability relationship network; designing a data inspection mechanism, and setting corresponding in-depth analysis conditions and trigger mechanisms for key parameters, respectively, implementing an in-depth analysis process to obtain in-depth analysis results; calculating comprehensive safety indicators based on dynamically adjusted risk assessment parameters; and executing warnings and recommended measures based on comprehensive safety indicators and in-depth analysis results. The present application obtains the potential risks of rope jumpers during rope skipping by performing a series of analyses on the video data of rope jumpers, and issues warnings and recommendations to rope jumpers, thereby achieving safe monitoring of rope skipping activities and effectively identifying and warning of potential safety risks.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of intelligent data recognition technology, and in particular to an intelligent rope skipping sports safety warning method and computer equipment based on behavior recognition. Background Art

[0002] Rope skipping is a simple and easy aerobic exercise that is widely popular among young students. However, rope skipping also has certain safety risks, such as improper posture leading to joint injuries and excessive fatigue leading to accidents. Traditional rope skipping safety education mainly relies on teachers' verbal guidance and demonstrations, and it is difficult to provide real-time and accurate guidance based on the individual differences of each student. In addition, in large-scale physical education classes or extracurricular activities, it is difficult for teachers to pay attention to the rope skipping movements of each student at the same time, and it is easy to ignore potential dangerous behaviors.

[0003] Existing methods for monitoring rope skipping safety mainly include manual observation and simple electronic counters. Manual observation is limited by the teacher's attention and experience, and it is difficult to continuously and comprehensively monitor the performance of each student. Although simple electronic counters can record the number of rope skipping, they cannot evaluate the quality and safety of rope skipping. Although some advanced sports monitoring devices can provide more detailed data, such as heart rate and calorie consumption, they are usually expensive, not suitable for large-scale use, and lack special optimization for rope skipping.

[0004] In recent years, computer vision and artificial intelligence technologies have made significant progress in the field of motion analysis. However, the application of these technologies in rope skipping safety monitoring is still limited. Most existing motion analysis systems are targeted at sports such as running and weightlifting, and are difficult to directly apply to high-frequency and complex movements such as rope skipping. In addition, these systems usually focus on performance improvement rather than safety monitoring, and cannot effectively identify and warn of potential safety risks. Summary of the invention

[0005] Based on this, it is necessary to provide an intelligent rope skipping sports safety warning method and computer equipment based on behavior recognition that can effectively monitor the safety of rope skipping in response to the above technical problems.

[0006] An intelligent rope skipping sports safety warning method based on behavior recognition comprises the following steps:

[0007] Collect and pre-process the video data of the rope jumper during the rope jumping activity; extract key features from the pre-processed data, and obtain the rope jumping status category based on the key features; the key features include body posture, rope jumping speed, rope jumping rhythm, relative positions of various body parts, motion trajectory and rope tension;

[0008] According to the key features, a data traceability relationship network is constructed; the data traceability relationship network is used to show the direct relationship and indirect impact path between the key features;

[0009] According to the rope skipping status category and data traceability relationship network, a data inspection mechanism is designed, and corresponding in-depth analysis conditions and trigger mechanisms are set for key parameters. When the in-depth analysis conditions and trigger mechanisms are met, the in-depth analysis process is implemented to obtain in-depth analysis results; key parameters are one or any combination of key features;

[0010] Define risk assessment parameters, initialize the risk assessment parameters based on the rope skipping status category, dynamically adjust the initialized risk assessment parameters based on the data inspection mechanism and the trigger mechanism, and calculate the comprehensive safety index based on the dynamically adjusted risk assessment parameters;

[0011] Implement early warning and recommended measures based on comprehensive safety indicators and in-depth analysis results.

[0012] In one embodiment, the step of collecting video data of a rope jumper during rope jumping activity and preprocessing the video data includes the following steps:

[0013] Collecting video data of a rope jumper; extracting multi-dimensional rope jumping data from the video data, the multi-dimensional rope jumping data including rope jumping action image data, rope jumping frequency data, rope jumper body shape data and rope tension data;

[0014] Simplify the multi-dimensional rope skipping data to obtain simplified data;

[0015] The simplified data is normalized to obtain normalized data.

[0016] In one embodiment, the step of extracting key features from the preprocessed data and obtaining the rope skipping status category according to the key features includes the steps of:

[0017] A dynamic time warping algorithm is used to compare the key features with features corresponding to the preset condition categories to obtain a similarity score corresponding to the key features;

[0018] Weight the similarity scores to obtain a comprehensive score;

[0019] Based on the combined score, the skipping rope condition category is identified.

[0020] In one embodiment, the step of constructing a data traceability relationship network according to the key features includes the following steps:

[0021] Calculate the cross-correlation function and conditional probability distribution between key features;

[0022] Based on multiple regression analysis, the dynamic relationship between key features is obtained;

[0023] Construct a weighted directed graph based on cross-correlation functions, conditional probability distributions, and dynamic relationships;

[0024] The edge weights of the weighted directed graph are calculated based on the maximum value of the cross-correlation function, the strength of the conditional probability distribution, and the regression coefficient of the dynamic relationship;

[0025] Based on weighted directed graphs and edge weights, a data traceability relationship network is constructed.

[0026] In one embodiment, the step of calculating the edge weight of the weighted directed graph based on the maximum value of the cross-correlation function, the strength of the conditional probability distribution and the regression coefficient of the dynamic relationship includes:

[0027] Determine the key factors for edge weight calculation, including the maximum value of the cross-correlation function, the strength of the conditional probability distribution, and the regression coefficient of the dynamic relationship;

[0028] Set the weight coefficients of key factors;

[0029] Based on key factors and weight coefficients, a comprehensive calculation model of edge weights is established;

[0030] The weight of each edge in the weighted directed graph is calculated using a comprehensive calculation model to obtain the edge weight of the weighted directed graph.

[0031] In one embodiment, a data inspection mechanism is designed according to the rope skipping status category and the data traceability relationship network, and corresponding in-depth analysis conditions and trigger mechanisms are set for key parameters respectively. When the in-depth analysis conditions and trigger mechanisms are met, the in-depth analysis process is implemented to obtain the in-depth analysis results, including the following steps:

[0032] Design basic tracking strategies based on rope skipping condition categories;

[0033] After designing the basic tracking strategy, determine the key parameters to be tracked based on the weighted directed graph corresponding to the data traceability relationship network;

[0034] After determining the key parameters, a data checking mechanism is designed based on the time delay relationship of the weighted directed graph;

[0035] After designing the data inspection mechanism, set corresponding in-depth analysis conditions and trigger mechanisms for key parameters;

[0036] When in-depth analysis conditions and trigger mechanisms are achieved, rapid backtracking checks, expanded inspection scope, cross-validation, contextual analysis and dynamic weight adjustments are implemented to achieve in-depth analysis results through the in-depth analysis process.

[0037] In one embodiment, when the in-depth analysis conditions and trigger mechanism are met, the steps of implementing rapid backtracking inspection, expanding the inspection scope, cross-validation, context analysis and dynamic weight adjustment to achieve the in-depth analysis process and obtain the in-depth analysis results include:

[0038] Perform quick back-tests to analyze changes in key parameters that trigger in-depth analysis over a predefined period of time;

[0039] When the quick back-check does not reveal obvious anomalies, the inspection scope is expanded to extend the analysis to a longer time period and more relevant parameters;

[0040] Conduct cross-validation, check the changes in other key parameters, and analyze the mutual influence between key parameters;

[0041] Perform contextual analysis by combining the jumper’s historical data and current training plan;

[0042] Based on the results of fast backtracking, expanded inspection scope, cross-validation, and context analysis, the weights are dynamically adjusted to update the edge weights in the weighted directed graph;

[0043] The results of rapid backtracking check, expanded check scope, cross-validation, context analysis and dynamic adjustment of weights are combined to obtain in-depth analysis results.

[0044] In one of the embodiments, risk assessment parameters are defined, the risk assessment parameters are initialized based on the rope skipping status category, the initialized risk assessment parameters are dynamically adjusted based on a data checking mechanism and a triggering mechanism, and a comprehensive safety index is calculated based on the dynamically adjusted risk assessment parameters;

[0045] Define risk assessment parameters; risk assessment parameters include movement safety, fatigue risk, technique stability, intensity appropriateness, and potential for improvement;

[0046] Initialize the safety of the action, fatigue risk, technical stability, appropriateness of intensity and potential for improvement based on the rope skipping condition category;

[0047] Based on the data inspection mechanism and trigger mechanism, the initialization of the rope skipping status category can dynamically adjust the action safety, fatigue risk, skill stability, intensity appropriateness and improvement potential;

[0048] The dynamically adjusted movement safety, fatigue risk, skill stability, intensity appropriateness and improvement potential are weighted to obtain a comprehensive safety index.

[0049] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0050] A computer-readable storage medium stores a computer program, which implements the steps of the above method when executed by a processor.

[0051] One of the above technical solutions has the following advantages and beneficial effects:

[0052] The present application is implemented through the following steps: collecting video data of a rope jumper during rope skipping activities and preprocessing it; extracting key features from the preprocessed data, and obtaining rope skipping status categories based on the key features; constructing a data traceability relationship network based on the key features; designing a data inspection mechanism based on the rope skipping status categories and the data traceability relationship network, and setting corresponding in-depth analysis conditions and trigger mechanisms for key parameters respectively, and when the in-depth analysis conditions and trigger mechanisms are achieved, implementing an in-depth analysis process to obtain in-depth analysis results; defining risk assessment parameters, initializing risk assessment parameters based on rope skipping status categories, dynamically adjusting the initialized risk assessment parameters based on the data inspection mechanism and the trigger mechanism, and calculating comprehensive safety indicators based on the dynamically adjusted risk assessment parameters; and executing early warning and recommended measures based on the comprehensive safety indicators and in-depth analysis results. The present application performs a series of analyses on the video data of rope jumpers to obtain the potential risks of rope jumpers during rope skipping, and provides early warnings and recommendations to rope jumpers, thereby achieving safe monitoring of rope skipping activities and effectively identifying and warning potential safety risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a flow chart of an intelligent rope skipping sports safety warning method based on behavior recognition in an embodiment of the present application.

[0054] Figure 2 Schematic diagram of the internal structure of a computer device in an embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0056] In order to achieve the above object, in one embodiment, Figure 1 As shown, a smart rope skipping sports safety warning method based on behavior recognition is provided, comprising the following steps:

[0057] Step S110, collecting and preprocessing video data of the rope jumper during the rope jumping activity; extracting key features from the preprocessed data, and obtaining the rope jumping status category according to the key features.

[0058] The system uses a high-frame rate, high-resolution camera to record at least 120 frames per second with a resolution of 4K to collect video data to capture every subtle change in the rope skipping activity. To obtain a full range of perspectives, the system uses multiple cameras to shoot from different angles, usually including front, side and top-down angles to ensure that the complete rope skipping action is captured.

[0059] In one example, the steps of collecting data of a rope jumper during a rope jumping activity and preprocessing the data include the following steps: collecting video data of the rope jumper; extracting multidimensional rope jumping data from the video data, the multidimensional rope jumping data including rope jumping action image data, rope jumping frequency data, rope jumper body data and rope tension data; streamlining the multidimensional rope jumping data to obtain simplified data; and normalizing the simplified data to obtain normalized data.

[0060] In one example, rope skipping action image data is obtained based on the following steps: a target detection algorithm is used to identify the positions of a rope jumper and a rope in video data; after identifying the positions of the rope jumper and the rope, an instance segmentation algorithm is used to segment the image of the rope jumper; the segmentation results in continuous video frames are analyzed to obtain the motion trajectories of each part of the rope jumper's identity, and an optical flow algorithm is used to calculate the rotation trajectory of the rope and the posture adjustment of the rope jumper; the motion trajectory, rotation trajectory and posture adjustment constitute the rope skipping action image data.

[0061] Specifically, the system first uses the YOLOv5 real-time target detection algorithm to identify the position of the jumper and the rope. YOLOv5 achieves fast and accurate target positioning by dividing the input image into grids and predicting multiple bounding boxes and category probabilities on each grid cell. The system specially trains the YOLOv5 model for the rope skipping scene, enabling it to accurately identify the jumper and the rope in complex backgrounds.

[0062] After recognition, the system uses the MaskR-CNN instance segmentation algorithm to perform instance segmentation, subdividing the body of the rope jumper into different parts, such as head, torso, limbs, etc. MaskR-CNN adds pixel-level segmentation on the basis of target detection, which can provide more detailed target contour information.

[0063] The system tracks the movement of the body parts of the jumper by analyzing the changes in these segmented areas in consecutive frames. At the same time, the system also uses optical flow algorithms (such as the Farneback algorithm) to calculate the pixel movement in the image sequence, which can capture more subtle movement changes, including the rotation trajectory of the jump rope and the jumper's slight posture adjustments (posture adjustments).

[0064] In one example, rope skipping frequency data is obtained based on the following steps: using an optical flow algorithm to track rope movement in video data; using time series analysis technology to obtain movement frequency from rope movement; calculating the number of mutations in the movement frequency per unit time to obtain rope skipping frequency data.

[0065] Specifically, the system uses the periodic characteristics of rope skipping to collect frequency data. First, the Lucas-Kanade optical flow algorithm is used to track the rope movement in the video data. The algorithm assumes that the pixel displacement between adjacent frames is small, and calculates the spatial and temporal derivatives of the image brightness in the local area to estimate the motion vector.

[0066] The system pays special attention to the moment when the rope passes under the feet of the jumper, which is usually manifested as a significant change in the optical flow field. In order to accurately capture this moment, the system also combines time series analysis techniques such as fast Fourier transform (FFT) to analyze the periodicity of the rope movement. FFT can convert time domain signals into frequency domain, making it easier to identify the main movement frequencies.

[0067] The system calculates the number of significant changes (mutation times) identified per unit time to obtain accurate rope skipping frequency data. In order to improve the stability of frequency measurement, the system also uses sliding window technology to calculate the average frequency in a shorter period of time (such as 3 seconds) and update this value in real time.

[0068] In one example, the body posture data of a rope jumper is obtained based on the following steps: using a human posture estimation algorithm to identify the key points of the rope jumper's body in video data; analyzing the relative positions and movements of the key points of the body to obtain the body posture data of the rope jumper.

[0069] Specifically, the system uses the OpenPose human pose estimation algorithm to identify and track the body key points of the rope jumper. OpenPose can detect up to 25 body key points at the same time, including the head, neck, shoulders, elbows, wrists, hips, knees, ankles, etc. It uses a method based on PartAffinityFields, which can not only detect a single key point, but also infer the connection relationship between key points.

[0070] The system obtains detailed body data of the rope jumper by analyzing the relative positions and movements of these key points of the body. For example, the system calculates the angles of each joint, such as the elbow angle, which can be obtained by calculating the angle between the vectors of the upper arm and forearm:

[0071] θ=arccos((v1·v2) / (|v1|*|v2|))

[0072] where v1 and v2 are the vectors of the upper arm and forearm respectively.

[0073] The system also calculates the velocity and acceleration of these body key points by comparing the changes in key point positions between consecutive frames:

[0074] v=(position_t-position_t-1) / Δt

[0075] a=(velocity_t-velocity_t-1) / Δt

[0076] Where v is the velocity, a is the acceleration, position_t is the position of the current frame, position_t-1 is the position of the previous frame, velocity_t is the velocity of the current frame, velocity_t-1 is the velocity of the previous frame, and Δt is the time interval between frames. To smooth the data and reduce the impact of noise, the system uses a Kalman filter to estimate velocity and acceleration.

[0077] In one example, rope tension data is obtained based on the following steps: an edge detection algorithm is used to identify the rope outline of the rope in the video data; a straight line detection technique is used to identify the straight line segment of the rope from the rope outline, and the curvature of the rope is obtained through the straight line segment; the video data is processed by an optical flow algorithm to obtain the movement speed of the rope; and a tension calculation model is used to process the curvature, movement speed, rope material and environmental factors to obtain rope tension data.

[0078] Specifically, the system first uses an edge detection algorithm to identify the rope outline of the rope in the video data. Through a series of image processing steps, the system is able to effectively extract the edges of the rope. Then, the system uses straight line detection technology to identify the straight segments of the rope. By analyzing the characteristics of these straight line segments, the system is able to quantify the degree of curvature of the rope. Specifically, the system calculates the deviation between the detected straight line segments and the ideal straight line to obtain the curvature of the rope. At the same time, the system also analyzes the movement speed of the rope, which can be achieved through the optical flow algorithm. The system establishes a simplified model to estimate the tension, which takes into account the curvature and speed of the rope. To improve the accuracy of the estimation, the system also considers the material properties of the rope and environmental factors (such as air resistance), and uses a machine learning model to optimize the tension estimation. This comprehensive approach enables the system to estimate the rope tension more accurately without direct measurement.

[0079] The rope skipping data is simplified to obtain simplified data, wherein the simplified data includes simplified rope skipping action image data, simplified rope skipping frequency data, simplified rope skipper posture data and simplified rope tension data.

[0080] In one example, the rope skipping action image data is simplified based on the following steps: continuous video frames are converted into a key frame sequence, and the degree of difference between adjacent frames in the key frame sequence is calculated, and the key frames with a degree of difference greater than a preset threshold are used as retained frames; differential coding technology is used to compress the rope skipping frequency data corresponding to the retained frame column to obtain rope skipping action image data.

[0081] Specifically, the system converts continuous video frames into a sequence of key frames. Specifically, the system selects a frame as a key frame every 0.1 seconds (or every 12 frames, based on a 120fps acquisition rate). This method can both preserve the continuity of the action and significantly reduce the amount of data. In each key frame, the system only retains the outline information of the jumper and the rope, and removes the background and other irrelevant elements. This operation is achieved by comparing the differences between adjacent frames, retaining only the areas that have changed.

[0082] The system focuses on retaining frames with significant changes in action, such as the moment of take-off and landing. Specifically, the system calculates the degree of difference between adjacent frames, and if the difference exceeds a preset threshold (for example, more than 30% of the pixels have changed), the frame is marked as a key frame and retained.

[0083] For the retained key frames, the system uses differential encoding technology for further compression. Specifically, the system only stores the differences between adjacent key frames instead of the complete frame content. This method takes advantage of the continuity of rope skipping and can significantly reduce data storage requirements.

[0084] In one example, the rope skipping frequency data is simplified based on the following steps: the rope skipping frequency data is processed using a sliding window technique to obtain the average frequency within the sliding window; a frequency mutation point is selected from the average frequency; and the frequency mutation point is encoded as a timestamp and a corresponding frequency value pair to obtain simplified rope skipping frequency data.

[0085] Specifically, the system uses a sliding window technique to calculate the average frequency every 0.5 seconds, rather than recording the exact time of each rope pass. This approach can smooth short-term fluctuations and provide a more stable frequency estimate. At the same time, the system only retains the time points where the frequency changes significantly (the change is more than 10%), thereby further reducing the amount of data.

[0086] The system focuses on the frequency mutation points. If the frequency changes by more than 20% in a short period of time (such as 0.5 seconds), the system will retain the data before and after this time point, and for periods of relatively stable frequency, the average value can be used instead to reduce the number of data points.

[0087] Frequency data is encoded as a series of timestamp and corresponding frequency value pairs. The system uses variable length encoding to store this information, using shorter codes for frequently occurring frequency values ​​to achieve data compression.

[0088] In one example, the body data of a rope jumper is simplified based on the following steps: the pixel coordinates of the key points of the body are converted into proportional coordinates relative to the height of the rope jumper; and based on the pixel coordinates of the key points of the body, the angle information of the joints of the rope jumper is calculated, and the angle information is encoded using quantization technology to obtain angle level information; the proportional coordinates and the angle level information constitute the simplified body data of the rope jumper.

[0089] Specifically, the system first simplifies the 25 key points of the body into the 15 most important points, including the head, neck, shoulders, elbows, wrists, hips, knees and ankles. For each key point, the system no longer stores the original pixel coordinates, but converts them into proportional coordinates relative to the height. This representation method makes the data more standardized and facilitates comparison between rope jumpers of different body shapes.

[0090] The system calculates and stores the angle information of key joints (such as elbows and knees) instead of the original position data. This representation more directly reflects the body posture while significantly reducing the amount of data.

[0091] The encoding of posture data uses quantization technology. The system discretizes the continuous angle values ​​into several predefined levels (such as 0-30 degrees, 30-60 degrees, etc.) to obtain angle level information, and each level is represented by a smaller bit. Although this method sacrifices some accuracy, it greatly reduces storage space, while still being accurate enough for security monitoring purposes.

[0092] Furthermore, the system focuses on retaining data on abnormal postures that may cause injury, such as excessive bending of the knees (angle less than 45 degrees) or excessive bending of the back (torso angle greater than 30 degrees from the vertical line). Posture data within the normal range can be appropriately simplified, for example, only retaining samples once per second.

[0093] In one example, the rope tension data is simplified based on the following steps: the rope tension data is divided into three tension levels according to the tension magnitude, and the tension level sequence and the time points corresponding to the level change positions are obtained; the tension level sequence and the time points corresponding to the level change positions are compressed using the run-length encoding technology to obtain the simplified rope tension data.

[0094] Specifically, the system no longer stores the rope tension data for each frame, but instead divides the data into three levels: high, medium, and low, and only records the time point when the level changes. This simplification not only retains important information about tension changes, but also significantly reduces the amount of data.

[0095] The system uses Run-Length Encoding to compress the tensile level data. For example, the continuous "Low-Low-Low-Medium-Medium-High" can be encoded as "3 Low 2 Medium 1 High", which greatly reduces the storage requirements.

[0096] The system also records the tension mutation points, that is, the moments when the tension level changes in a short period of time (such as 0.2 seconds). These mutation points are crucial for analyzing the stability and potential risks of rope skipping movements.

[0097] After these processing steps, the raw data are integrated into a unified and simplified dataset of rope skipping activities. The data are stored in a compact binary format, with each rope skipping cycle (a complete jump) as a basic unit.

[0098] The rope skipping activity data is normalized to obtain normalized data, wherein the normalized data includes normalized rope skipping action image data, normalized rope skipping frequency data, normalized rope skipper posture data and normalized rope tension data.

[0099] First, for the rope skipping action image data, the system uses pixel value standardization processing. Specifically, the system linearly maps the RGB value of each pixel from the range of 0-255 to the range of 0-1. This processing not only reduces the storage requirements of the data, but also facilitates subsequent image processing and analysis. At the same time, in order to further enhance the contrast of the image and highlight key features, the system also applies histogram equalization technology. This technology redistributes the grayscale value distribution of the image to make the brightness distribution of the overall image more uniform, which is conducive to identifying subtle changes in rope skipping movements. In addition, the system also denoises the image, using a median filter to remove random noise that affects the analysis while retaining the edge information of the image. These processing steps ensure that the action image data is more suitable for subsequent computer vision analysis while retaining key information.

[0100] For the rope skipping frequency data, the system uses the minimum-maximum normalization method to map it to the range of 0 to 1. Considering that the normal rope skipping frequency is usually between 60 and 200 times / minute, the system sets 60 times / minute as the minimum value of 0 and 200 times / minute as the maximum value of 1, and then linearly maps the actual frequency according to this range. For extreme values ​​outside this range, the system uses a soft threshold processing method, which allows values ​​to slightly exceed the 0-1 range, but applies a fast decaying function to limit its impact. This processing method not only retains the relative relationship of the frequency data, but also can handle outliers, while making the frequency data easier to analyze in combination with other normalized data. In addition, the system also calculates the moving average and standard deviation of the frequency. These derived features help capture the changing trend and stability of the frequency.

[0101] The normalization of the body data of rope jumpers includes two main aspects: key point coordinates and joint angles. For the coordinate data of 15 key points, the system first converts the original pixel coordinates into proportional coordinates relative to the height, and then further maps these proportional values ​​to the range of [-1,1]. This double conversion not only eliminates the differences between rope jumpers of different heights, but also makes the data more suitable for the input of machine learning models. For joint angle data, the system maps the original 0 to 360 degree angle values ​​to the range of -1 to 1. This conversion is achieved by dividing the angle value by 180 degrees and then subtracting 1, so that 90 degrees are converted to 0, 180 degrees are converted to 1, and 0 degrees are converted to -1. This normalization not only makes the angle data more intuitive, but also more suitable for use in various machine learning models. At the same time, the system also calculates the joint angular velocity and angular acceleration, and these derived features are crucial for analyzing the smoothness and coordination of movements.

[0102] Finally, for the rope tension data, the system has simplified it into three levels: high, medium, and low in step S120. In the normalization process, the system further converts these three discrete levels into continuous numerical values. Specifically, the system maps the low level to 0.167, the medium level to 0.5, and the high level to 0.833. This conversion allows the tension data to be numerically analyzed together with other continuous data. At the same time, the system also introduces the derived feature of tension change rate, which quantifies the dynamic changes of tension by calculating the difference in tension values ​​at adjacent time points. In addition, the system also records the maximum, minimum, and average tension in each rope skipping cycle. These statistical features help to fully understand the change pattern of tension throughout the rope skipping process.

[0103] Through this comprehensive and detailed normalization process, all data types in the simplified rope skipping activity dataset have been converted into a unified, directly comparable format. This normalization not only allows data from different sources and different types to be directly compared and comprehensively analyzed, but also greatly improves the scalability and flexibility of the system. When new rope skipping skills or analysis tasks need to be added, the new data can be seamlessly integrated into the existing system by simply processing it according to this set of normalization methods.

[0104] Key features are extracted from the normalized data, and the rope skipping status categories are obtained based on the key features. The key features include body posture, rope skipping speed, rope skipping rhythm, relative positions of various body parts, movement trajectory, and rope tension; the rope skipping status categories are excellent, good, average, needs improvement, or potential risk.

[0105] Specifically, 1. Extract key features: Body posture P(t): Based on the normalized posture data and motion image data, the system extracts the three-dimensional coordinates of 15 key points (head, neck, shoulders, elbows, wrists, torso center, hips, knees and ankles). Use a sliding window of 5 frames (about 0.04 seconds) to calculate the velocity and acceleration of each key point. Direction vectors of key bone chains (spine, upper arm, forearm, thigh, calf) are directly obtained from the posture data, and the angles between adjacent bone chains are calculated. These data form a 450-dimensional vector (15 points * 3 coordinates * 3 types of data (position, velocity, acceleration) + 10 bone chain angles), which is updated every 20ms to form a time series P(t).

[0106] Rope skipping speed V(t) and rope skipping rhythm R(t): Based on the normalized frequency data, the system records the number of rope skipping per second and measures the time interval between consecutive jumps. The data of the last 10 seconds is subjected to a fast Fourier transform to extract the main rope skipping frequencies and their amplitudes from the frequency domain. The mean, standard deviation, and coefficient of variation of the speed are calculated in the time domain. For rhythm analysis, the autocorrelation function of the time interval series is calculated, and the position and amplitude of the first and second peaks are extracted. These features are combined into a 10-dimensional vector, which is updated once per second. Among them, the features directly related to the speed constitute the V(t) time series, and the features related to the rhythm form the R(t) time series.

[0107] Relative position of each part of the body RP(t): Based on the normalized posture data and motion image data, a skeletal topology structure containing 15 nodes is constructed. The Euclidean distance between adjacent joints and its rate of change over time are calculated, and the angle between the trunk and the limbs and their angular velocity are calculated. These data form a 105-dimensional vector (15 node pairwise distance combination * 2 (distance and change rate) + 8 main joint angles and their angular velocities), which is updated every 20ms to form a time series RP(t).

[0108] Motion trajectory MT(t): Based on the normalized motion image data and body posture data, cubic spline interpolation is performed on the three-dimensional coordinate sequence of each key point to generate a continuous motion curve. The tangent vector, normal vector and binormal vector of the curve at each time point are calculated, and the curvature and torsion are calculated based on these vectors. For the wrist and ankle points, the normal vector of their motion plane is additionally calculated. These features form a 225-dimensional vector (15 points * 3 class vectors * 3 coordinates + 15 points * 2 (curvature and torsion) + 4 special points of the plane normal vector * 3 coordinates), which is updated every 20ms to form a time series MT(t).

[0109] Rope tension T(t): Based on the normalized tension data and action image data, the shape of the rope is identified through image processing technology and approximated as multiple Bezier curves. The tension value is accurately estimated using a physical model by combining the curvature of the curve, the acceleration of the rope jumper's wrist, and the directly measured tension data. The mean, standard deviation, peak, and valley values ​​of the tension are calculated, and the tension sequence is subjected to wavelet transformation to extract the energy distribution at different scales. These features form a 20-dimensional vector that is updated every 50ms to form a time series T(t).

[0110] In one example, the steps of extracting key features from normalized data and obtaining rope skipping condition categories based on the key features include: using a dynamic time warping algorithm to compare the key features with features corresponding to preset condition categories to obtain similarity scores corresponding to the key features; weighting the similarity scores to obtain comprehensive scores; and identifying rope skipping condition categories based on the comprehensive scores.

[0111] After extracting key features, the system uses the dynamic time warping (DTW) algorithm to compare the current rope skipping features with the features of the preset status categories. The system presets five rope skipping status categories: excellent, good, average, needs improvement, and potential risk. Each category has its own specific feature description, expressed as:

[0112] S_i={P_i(t),V_i(t),R_i(t),RP_i(t),MT_i(t),T_i(t)}

[0113] Where i represents the category index (1=excellent, 2=good, 3=average, 4=needs improvement, 5=potential risk), and t represents the time series.

[0114] The system uses the dynamic time warping (DTW) algorithm to compare the current features of each skipping rope with the features of each preset condition category to obtain a series of similarity scores. Taking posture features as an example, the DTW calculation process is as follows:

[0115] 1. Create a distance matrix D: D(m,n)=||P_current(m)-P_i(n)|| where P_current is the current posture data, P_i is the standard posture data of the i-th condition category, and m and n are the time indexes of the current data and the standard data, respectively.

[0116] 2. Calculate the cumulative distance matrix C: C(m,n)=D(m,n)+min(C(m-1,n),C(m,n-1),C(m-1,n-1))

[0117] 3. The final DTW distance is the lower right corner element of the C matrix: DTW_posture(i)=C(M,N) where M and N are the total lengths of the current data and the standard data, respectively.

[0118] Similar DTW calculations are performed on other features (speed, rhythm, relative position, motion trajectory, rope tension) to obtain DTW_velocity(i), DTW_rhythm(i), DTW_position(i), DTW_trajectory(i) and DTW_tension(i).

[0119] The system then uses a composite scoring formula to calculate a score for each condition category:

[0120] Score(i)=w1*(1 / DTW_posture(i))+w2*(1 / DTW_velocity(i))+w3*(1 / DTW_rhythm(i))+w4*(1 / DTW_position(i))+w5*(1 / DTW_trajectory(i))+w6*(1 / DTW_tension(i))

[0121] Where w1 to w6 are weight coefficients. The skipping conditions are divided into the following five categories based on the scoring system:

[0122] 1. Excellent: Score ≥ 0.90 means that the rope jumper's movements are standardized, the speed and rhythm are stable, the posture is coordinated, and all indicators have reached a high level.

[0123] 2. Good: 0.80≤Score<0.90 indicates that the overall performance of the rope jumper is good, the movements are basically correct, the speed and rhythm are relatively stable, but there is still room for improvement in some details.

[0124] 3. Average: 0.70≤Score<0.80 indicates that the skipper’s performance is at an average level, with small deviations in movements, slight fluctuations in speed or rhythm, and average coordination of posture.

[0125] 4. Needs improvement: 0.60≤Score<0.70 indicates that there are obvious problems in the performance of the rope jumper, with large deviations in movements, unstable speed or rhythm, and poor posture coordination.

[0126] 5. Potential risks: Score < 0.60 means that the jumper's movements are seriously deviated from the standard, the speed or rhythm is extremely unstable, and the posture may cause injury, which requires immediate adjustment.

[0127] Step S120: construct a data traceability relationship network based on the key features, wherein the data traceability relationship network is used to display the direct relationship and indirect impact path between the key features.

[0128] In one example, the steps of constructing a data traceability relationship network based on key features include: calculating the cross-correlation function between body posture, rope skipping speed, rope skipping rhythm, relative positions of various body parts, motion trajectory and rope tension; calculating the conditional probability distribution between body posture, rope skipping speed, rope skipping rhythm, relative positions of various body parts, motion trajectory and rope tension; obtaining the dynamic relationship between body posture, rope skipping speed, rope skipping rhythm, relative positions of various body parts, motion trajectory and rope tension based on multiple regression analysis; constructing a weighted directed graph based on the cross-correlation function, conditional probability distribution and dynamic relationship; calculating the edge weight of the weighted directed graph based on the maximum value of the cross-correlation function, the intensity of the conditional probability distribution and the regression coefficient of the dynamic relationship; constructing a data traceability relationship network based on the weighted directed graph and the edge weights.

[0129] The system first calculates the cross-correlation function between the six key features extracted in step S110 (body posture P(t), rope skipping speed V(t), rope skipping rhythm R(t), relative position of each body part RP(t), motion trajectory MT(t) and rope tension T(t)) to determine the precise time delay between the key features. The cross-correlation function R_xy(τ) is defined as: , where E is the expected value, μ is the mean, and σ is the standard deviation. By analyzing the data of 1,000 skipping cycles, the system discovered some key time delay relationships. For example:

[0130] -Between body posture P(t) and relative positions of body parts RP(t): P(t) usually leads RP(t) by about 0.1 seconds, and the cross-correlation function between the two reaches a peak of 0.75 at this time. This means that if the body posture of the jumper changes at t=1.0 seconds (such as an increase in torso angle of 5 degrees), the system may observe a corresponding change in the relative positions of body parts at t=1.1 seconds.

[0131] - Relative position of body parts RP(t) and skipping speed V(t): RP(t) changes usually lead V(t) by about 0.2 seconds, with a peak correlation coefficient of 0.68. For example, a 10-degree change in the angle of the jumper's arm position relative to the torso at t=2.0 seconds may result in a speed change at t=2.2 seconds.

[0132] -Between skipping speed V(t) and skipping rhythm R(t): Changes in V(t) usually precede R(t) by about 0.1 seconds, with a peak correlation coefficient of 0.82. For example, a sudden change in the speed of a skipper at t=3.0 seconds may lead to a corresponding adjustment in the rhythm at t=3.1 seconds.

[0133] - Between the motion trajectory MT(t) and the rope tension T(t): The change of MT(t) usually leads T(t) by about 0.05 seconds, with a peak correlation coefficient of 0.82. For example, when the jumper's wrist trajectory suddenly accelerates at t=4.0 seconds, it may cause an increase in the rope tension at t=4.05 seconds.

[0134] Next, the system analyzes the conditional probability distribution of other key features under the condition of specific key feature values. The system introduces conditional probability P(Y|X) to express the probability that parameter Y falls into a certain range given a specific range of values ​​of parameter X. For example, for the relationship between the relative position RP(t) of various parts of the body and the skipping speed V(t), the system calculates P(V|RP), that is, the probability distribution of V(t) under the condition of given RP(t). The system found that when the relative position of the hip joint and the shoulder joint is maintained within a specific range (such as the hip-shoulder distance is 45%-50% of the height), the probability of the skipping speed V(t) remaining in the ideal range (such as 120-140 times per minute) increases to 80%. Expressed as conditional probability: P(V in the ideal range | RP in the specific range) = 0.8. Similarly, for the relationship between body posture P(t) and rhythm R(t), the system calculates P(R|P) and finds that when the trunk remains vertical (the angle with the ground is between 85°-95°), the probability that the skipping rhythm R(t) remains stable (the change in the interval between adjacent jumps is less than 5%) reaches 75%. The conditional probability is expressed as: P(R remains stable|P remains vertical) = 0.75.

[0135] In order to quantify the dynamic relationship between key features, the system uses multivariate regression analysis. The system first collects a large amount of historical data, including time series of various parameters. These data cover multiple skipping processes of rope jumpers with different skill levels, ensuring the representativeness and diversity of the sample. Then, the system segments these data into predefined time windows (such as 1 second) and calculates the average value of each parameter in each window. This processing can reduce the impact of noise and improve the stability of the data. Based on these processed data, the system establishes multiple regression models to describe the relationship between parameters. For example, the system establishes a relationship model between body posture P(t), relative position RP(t) of various parts of the body and rope skipping speed V(t): V(t)=β0+β1*P_trunk(t)+β2*RP_arm(t)+ε, where P_trunk represents the trunk angle, RP_arm represents the angle of the arm relative to the trunk, β0 is the intercept, β1 and β2 are regression coefficients, and ε is the error term. Another example is the relationship model between the motion trajectory MT(t) and the rope tension T(t): T(t)=β0+β1*MT_y(t)+β2*MT_y(t-0.1)+ε, where MT_y represents the vertical position of the wrist, β0 is the intercept, β1 and β2 are regression coefficients, and ε is the error term.

[0136] The system estimates the regression coefficients using the least squares method and determines the optimal β value by minimizing the sum of squared errors between the predicted values ​​and the actual values. Specifically, the system performs the following steps for each model: First, calculate the residual sum of squares (RSS): , where y_i is the actual observed value, is the model prediction. Then, the best β value is found by minimizing RSS. This process usually involves solving the system of equations by taking the partial derivative of RSS with respect to β and setting it to zero. In practice, the system may use numerical optimization methods, such as gradient descent or Newton's method, to find the optimal solution. For the first model mentioned above, the system may obtain β0=120, β1=2, and β2=-1.5. This means that for every 1 degree increase in the torso angle, the speed increases by 2 times / minute; and for every 1 degree increase in the arm angle relative to the torso, the speed decreases by 1.5 times / minute. These regression coefficients not only provide a quantitative description of the relationship between the parameters, but also reveal the relative importance of each factor to the speed of skipping rope.

[0137] After completing the cross-correlation analysis, conditional probability calculation, and regression analysis, the system begins to construct a weighted directed graph G=(V,E), where the vertex V represents six key features (body posture P(t), skipping speed V(t), skipping rhythm R(t), relative position of each body part RP(t), motion trajectory MT(t), and rope tension T(t)), and the edge E represents the influence relationship between the features. The construction process of this graph involves multiple steps, each of which is carefully designed to ensure that the final graph can accurately reflect the complex relationship between the parameters. The system first determines the direction of the edge in the graph based on the results of the cross-correlation analysis. This step is based on the time delay relationship. If the change of parameter X usually leads the change of parameter Y, there will be a directed edge from X to Y. For example, based on the previous analysis, the system found that body posture P(t) usually leads speed V(t) by about 0.18 seconds, so there will be an edge from P to V. Similarly, the system determines the direction of all edges in the graph based on the time delay relationship between other parameters.

[0138] Next, the system calculates the weight of each edge. This weight value takes into account three key factors: the maximum value of the cross-correlation function, the strength of the conditional probability distribution, and the coefficients obtained from the regression analysis. The system uses a weighted summation formula to combine these three factors to obtain the final edge weight. Specifically, for the edge from parameter X to parameter Y, its weight can be expressed as: Weight=α*max(R_xy(τ))+β*P(Y|X)+γ*(|β1|+|β2|) / 2

[0139] in:

[0140] -max(R_xy(τ)) is the maximum value of the cross-correlation function between X and Y

[0141] -P(Y|X) is the conditional probability calculated previously

[0142] -β1 and β2 are the coefficients obtained from the regression analysis

[0143] -α, β, and γ are adjustment factors used to balance the relative importance of the three components

[0144] In this way, the system calculates the weights of each edge: The edge weight from P(t) to V(t) is 0.75, reflecting the strong correlation between them. The edge weight from RP(t) to R(t) is 0.68, indicating that the relative position of each body part has a significant effect on the rhythm. The edge weight from MT(t) to T(t) is 0.82, indicating that the movement trajectory has a great influence on the rope tension.

[0145] In this graph structure, the system analyzes the influence paths between key features. The system observes a significant influence path: P(t)→RP(t+0.1)→V(t+0.3)→R(t+0.4), indicating that changes in body posture are associated with changes in speed and rhythm by affecting the relative positions of various parts of the body. Specifically:

[0146] 1. The change in the body posture of the rope jumper, P(t), is correlated with the change in the relative position of each body part, RP(t+0.1), after 0.1 seconds, with a correlation strength of 0.75.

[0147] 2. This in turn correlates with the change in rope skipping speed V(t+0.3) 0.2 seconds later, with a correlation strength of 0.68.

[0148] 3. Finally, it was correlated with the change of rope skipping rhythm R(t+0.4) after 0.1 seconds, with a correlation strength of 0.82.

[0149] In this way, the system builds a detailed data traceability network. This network not only shows the direct relationship between parameters, but also reveals complex indirect influence paths. For example:

[0150] 1. The change of body posture P(t) usually precedes the change of speed V(t+0.3) and rhythm R(t+0.4), indicating that body posture may be a key factor affecting speed and rhythm.

[0151] 2. The change in rope tension T(t) slightly precedes the change in speed V(t+0.1), reflecting the potential relationship between the jumper's ability to control the rope and the speed of skipping.

[0152] 3. The relative position RP(t) of each body part is closely related to the velocity V(t), and its change is usually associated with the change of velocity V(t+0.2) after 0.2 seconds.

[0153] 4. The motion trajectory MT(t) plays a key role in connecting various parameters in the entire network by affecting the body posture P(t) and directly acting on the velocity V(t).

[0154] This detailed data traceability relationship is closely integrated with the status classification result in step S110. For example, if step 110 determines that the current rope skipping status is "needs improvement", the system can track the key factors that lead to this status through the traceability relationship network. It may find that the abnormal body posture P(t) (such as the trunk leaning forward more than 10 degrees) causes the subsequent speed V(t) to drop and the rhythm R(t) to be unstable, thereby providing more accurate improvement suggestions for the rope jumper, such as "keep the trunk straight and avoid leaning forward too much".

[0155] Step S130, design a data inspection mechanism according to the rope skipping status category and the data traceability relationship network, and set corresponding in-depth analysis conditions and trigger mechanisms for key parameters, and implement the in-depth analysis process to obtain in-depth analysis results when the in-depth analysis conditions and trigger mechanisms are met. Among them, the key parameters are one or any combination of the key features.

[0156] In an example, a data checking mechanism is designed according to the rope skipping status category and the data traceability relationship network, and corresponding in-depth analysis conditions and trigger mechanisms are set for key parameters. When the in-depth analysis conditions and trigger mechanisms are met, the steps of implementing the in-depth analysis process to obtain in-depth analysis results include the following steps: designing a basic tracking strategy according to the rope skipping status category; after designing the basic tracking strategy, determining the key parameters to be tracked according to the weighted directed graph corresponding to the data traceability relationship network; after determining the key parameters, designing a data checking mechanism based on the time delay relationship of the weighted directed graph; after designing the data checking mechanism, setting corresponding in-depth analysis conditions and trigger mechanisms for key parameters; when the in-depth analysis conditions and trigger mechanisms are met, implementing rapid backtracking checks, expanding the scope of checks, cross-validation, context analysis, and dynamic weight adjustment to implement the in-depth analysis process to obtain in-depth analysis results.

[0157] 1. Basic tracking strategy design:

[0158] The design process first designs a basic tracking strategy using the five status categories (excellent, good, average, needs improvement, and potential risk) determined in step S110. For each category, the system sets different data sampling intervals and parameter warning thresholds. For example, for the "excellent" category (score ≥ 0.90), the system uses a longer sampling interval, such as 0.5 seconds, focusing on factors that may cause slight performance declines. In contrast, for the "potential risk" category (score < 0.60), the system uses high-frequency sampling, such as once every 0.05 seconds, and pays close attention to any abnormal changes. This differentiated sampling strategy ensures that the system can provide appropriate monitoring density based on the overall level of the rope jumper.

[0159] 2. Determination of key parameters:

[0160] After setting the basic tracking strategy, the system uses the weighted directed graph constructed in step S120 to determine the key parameters and their relationships that need to be tracked. The system selects the rope skipping speed data V(t) as the core starting point for tracking because speed is a direct reflection of rope skipping performance and the result of the combined effect of multiple other parameters. Based on the weighted directed graph, the system determines the parameters that need to be focused on: posture data P(t), rope tension data T(t) and rhythm data R(t). The system also directly considers the influence of the relative position RP(t) of various parts of the body and the motion trajectory MT(t) on the speed.

[0161] 3. Data inspection mechanism design:

[0162] After determining the key parameters, the system designed a data checking mechanism based on the time delay relationship marked in the weighted directed graph. This mechanism determines when the system checks the data of which parameters. For example, if a speed anomaly is observed at time t, the system will check the relevant parameters in the following order: check the P(t) data at t-0.18 seconds, check the T(t) data at t-0.05 seconds, check the R(t) data at t+0.1 seconds, check the RP(t) data at t-0.2 seconds, and check the MT(t) data at t-0.05 seconds. This time delay-based checking mechanism enables the system to accurately capture the causal relationship between parameter changes.

[0163] 4. Analysis conditions and trigger mechanism settings:

[0164] After establishing the data inspection mechanism, the system sets specific analysis conditions and trigger mechanisms for each key parameter. The setting of these conditions and mechanisms is directly related to the system's ability and efficiency in identifying potential problems.

[0165] For the rope skipping speed V(t), the system sets multiple levels of trigger conditions: instantaneous trigger (single jump speed abnormality), short-term trigger (multiple consecutive jump speed abnormalities) and trend trigger (speed shows a continuous downward trend). Specifically, for the "excellent" category, if the speed of three consecutive sampling points drops by more than 10%, an in-depth analysis is triggered; and for the "potential risk" category, as long as the speed of a single sampling point drops by more than 5%, an in-depth analysis is immediately triggered.

[0166] For body posture P(t), the system sets the following analysis conditions: angle change exceeds a threshold (the threshold is between 5 and 15 degrees, depending on the condition category); posture duration exceeds a set value (for example, the trunk leans forward more than 10 degrees and lasts for more than 1 second); and posture change rate is abnormal (such as an angle change of more than 20 degrees within 0.1 seconds).

[0167] For the rope tension T(t), the trigger conditions include: the instantaneous tension exceeds the preset threshold (such as 20% of body weight); the tension changes suddenly (such as a change of more than 50% within 0.1 second); the tension is abnormally continuous (such as the tension exceeds the normal range for 5 consecutive jumps).

[0168] The analysis conditions of the skipping rope rhythm R(t) include: sudden change of rhythm (such as the difference between the interval time of two consecutive jumps exceeds 30%); continuous instability of rhythm (such as the coefficient of variation of the interval time of 10 consecutive jumps exceeds 20%); mismatch between rhythm and speed (such as the rhythm speeds up but the speed decreases).

[0169] The system also sets corresponding analysis conditions for the relative position RP(t) and motion trajectory MT(t) of various body parts that directly affect the speed. The trigger conditions of the relative position RP(t) of various body parts include: key joint angles outside the normal range (such as elbow flexion angle less than 90 degrees); abnormal relative position of body parts (such as wrist position lower than hip); asymmetric posture (such as left and right arm height difference of more than 10 cm). The analysis conditions of the motion trajectory MT(t) include: abnormal jump height (such as 3 consecutive jumps with a height drop of more than 20%); irregular motion trajectory (such as a sudden increase in the left and right swing amplitude); change in landing mode (such as changing from forefoot landing to full foot landing).

[0170] The sensitivity of these trigger mechanisms is dynamically adjusted according to the jumper's condition category. For example, for jumpers in the "excellent" category, the system will slightly increase the trigger threshold to avoid too many false alarms; while for jumpers in the "potential risk" category, the system will lower the trigger threshold to ensure that any potential problems can be caught in time.

[0171] When setting these analysis conditions and trigger mechanisms, the system takes into account the weight information in the weighted directed graph. For example, the edge weight from P(t) to V(t) is 0.75, the edge weight from T(t) to V(t) is 0.82, the edge weight from RP(t) to V(t) is 0.68, and the edge weight from MT(t) to V(t) is 0.82. Based on this, the system calculates the contribution of each parameter to the observed speed anomaly and adjusts the sensitivity of the trigger conditions accordingly.

[0172] 5. Composite trigger condition design:

[0173] In order to more comprehensively analyze abnormal situations during rope skipping, the system also designs composite trigger conditions. These conditions take into account the interaction between multiple parameters and can capture complex situations that may be overlooked by single parameter analysis. For example, if the system finds that changes in body posture (P(t) abnormality) always occur before a decrease in rope skipping speed (V(t) abnormality), it will set a composite trigger condition: when a significant change in body posture is detected, the analysis process will be triggered in advance even if the speed has not yet decreased. Specifically, if the system records that in 90% of cases, an increase in the trunk angle of more than 8 degrees will cause a speed decrease of more than 10% after 0.15 seconds to 0.2 seconds, it will set an early warning trigger: as long as an increase in the trunk angle of 8 degrees is detected, the analysis process will be started without waiting for the speed to actually decrease.

[0174] 6. In-depth analysis process:

[0175] When these single or combined conditions are triggered, the system initiates an in-depth analysis process, which includes rapid backtracking, expanded inspection scope, cross-validation, contextual analysis, and dynamic weight adjustment.

[0176] In the fast backtracking check, the system immediately checks all relevant parameter data within the past 1 second to look for possible triggering factors. If no obvious cause is found within 1 second, the system expands the check scope to the past 5 seconds, which constitutes the expanded check scope step.

[0177] During the cross-validation phase, the system checks whether other parameters also show abnormal changes. For example, if the posture data P(t) is found to be abnormal, the system will further analyze the change sequence of P(t) and V(t), determine the causal relationship between them, and possibly set new compound trigger conditions.

[0178] In contextual analysis, the system considers broader contextual information, such as the jumper’s historical performance, the day’s training plan, etc., and adjusts the trigger conditions accordingly. For example, if the system knows that the jumper is doing high-intensity interval training, then a sharp drop in speed observed during the rest phase will not trigger a deeper analysis.

[0179] Finally, in the dynamic weight adjustment stage, the system temporarily adjusts the influence weights between certain parameters and the corresponding trigger conditions based on the analysis results. For example, if it is found that the change in posture of this particular jumper has a particularly significant impact on speed, the system may temporarily increase the edge weight from P(t) to V(t) from 0.75 to 0.85, while lowering the trigger threshold associated with P(t).

[0180] This in-depth analysis results will be used to optimize the risk assessment process, such as adjusting the risk assessment parameter weights and thresholds to make the risk assessment more accurate. At the same time, these results will also be used to improve the warning trigger mechanism. The system may create more accurate warning conditions based on the newly discovered relationship between parameters. In addition, the accumulated analysis results will be used to identify the long-term performance patterns of rope jumpers, which will help predict potential risks and design long-term improvement strategies.

[0181] Based on the above design, the system may draw the following analysis conclusion: the jumper's arm position dropped due to fatigue at 24.62 seconds, triggering the analysis conditions related to RP(t). This triggered a series of chain reactions: the trunk leaned forward compensatorily (24.82 seconds), triggering the analysis conditions of P(t); then the rope tension increased (24.95 seconds), triggering the analysis conditions of T(t); and finally at 25 seconds, it showed a significant decrease in speed, triggering the main analysis conditions of V(t). Each step of the triggered analysis provides the system with more information and helps to build a complete problem evolution process. This process not only reflects the impact of fatigue accumulation on rope skipping performance, but also ensures that problems can be discovered and analyzed in a timely manner through multiple trigger mechanisms.

[0182] Step S140, defining risk assessment parameters, initializing the risk assessment parameters based on the rope skipping status category, dynamically adjusting the initialized risk assessment parameters based on the data checking mechanism and the triggering mechanism, and calculating the comprehensive safety index based on the dynamically adjusted risk assessment parameters.

[0183] In an example, risk assessment parameters are defined, the risk assessment parameters are initialized based on the rope skipping status category, the initialized risk assessment parameters are dynamically adjusted based on a data checking mechanism and a triggering mechanism, and a comprehensive safety index is calculated based on the dynamically adjusted risk assessment parameters; risk assessment parameters are defined; the risk assessment parameters include movement safety, fatigue risk, skill stability, intensity appropriateness and improvement potential; based on the rope skipping status category, movement safety, fatigue risk, skill stability, intensity appropriateness and improvement potential are initialized; based on the data checking mechanism and the triggering mechanism, the movement safety, fatigue risk, skill stability, intensity appropriateness and improvement potential are dynamically adjusted for the initialized rope skipping status category; the dynamically adjusted movement safety, fatigue risk, skill stability, intensity appropriateness and improvement potential are weighted to obtain a comprehensive safety index.

[0184] First, the system defines five main risk assessment parameters: action safety (S), fatigue risk (F), skill stability (T), intensity appropriateness (I), and improvement potential (P). Action safety (S) evaluates the standardization of rope skipping movements. Fatigue risk (F) monitors the fatigue level of rope jumpers to prevent overtraining. Skill stability (T) measures the proficiency and consistency of rope skipping techniques. Intensity appropriateness (I) determines whether the current rope skipping intensity is appropriate. Improvement potential (P) considers the long-term development of rope jumpers and evaluates them based on historical data and current performance.

[0185] The five risk assessment parameters are initialized based on the five status categories (excellent, good, average, needs improvement, and potential risk) determined in step S110. The initial value of each parameter is closely related to the status category of the rope jumper, ensuring that the starting point of the risk assessment matches the overall performance level of the rope jumper. For rope jumpers in the "excellent" category, the system assigns higher initial values ​​to action safety (S) and skill stability (T), such as 0.9; assigns lower initial values ​​to fatigue risk (F), such as 0.2; and assigns medium-to-high initial values ​​to intensity appropriateness (I) and potential for improvement (P), such as 0.8. This reflects that excellent rope jumpers usually have good movement control ability, lower fatigue risk, and higher training adaptability and room for improvement.

[0186] On the contrary, for the "potential risk" category of rope jumpers, the system will assign lower initial values, such as 0.5, to the safety of action (S) and the stability of technique (T); assign higher initial values, such as 0.7, to the risk of fatigue (F); and assign lower initial values, such as 0.4, to the appropriateness of intensity (I) and the potential for improvement (P). This setting reflects that this type of rope jumper faces higher safety risks, their movements are not standardized, they are prone to fatigue, and their training intensity is not appropriate. They have more room for improvement but also face more challenges.

[0187] For the three intermediate categories of "good", "average" and "needs improvement", the system assigns corresponding intermediate values ​​to the five risk factors according to their relative positions in the five categories. For example, the initial values ​​of the "good" category are: S=0.8, F=0.3, T=0.8, I=0.75, P=0.7; the "average" category is: S=0.7, F=0.5, T=0.7, I=0.6, P=0.6; the "needs improvement" category is: S=0.6, F=0.6, T=0.6, I=0.5, P=0.5. This gradient initial value setting ensures that the system can accurately reflect the initial safety status of rope jumpers of different levels.

[0188] After initialization, the system enters the dynamic adjustment phase. This phase uses the reverse tracking path and trigger mechanism designed in step S130 to adjust the risk assessment parameters based on real-time feedback. When the system detects that a trigger condition is activated, it will adjust one or more risk assessment parameters accordingly based on the nature and severity of the trigger condition. For example, if the speed data V(t) triggers a short-term trigger condition, such as the average speed of three consecutive jumps is less than 85% of the normal level, the system will reduce the value of the intensity appropriateness (I) parameter. The magnitude of the reduction is proportional to the severity of the trigger condition and will reduce the I value by 0.1 or more.

[0189] Similarly, if the posture data P(t) triggers a posture duration condition, such as a forward lean of the trunk exceeding 15 degrees for more than 1 second, the system reduces the value of the action safety (S) parameter. This reduction is more significant because incorrect posture directly leads to injury. The system reduces the S value by 0.2 or more, depending on the degree and duration of the posture deviation. At the same time, the system also slightly increases the value of fatigue risk (F) because incorrect posture is often associated with fatigue.

[0190] The skill stability (T) parameter is mainly affected by the rhythm data R(t) and the motion trajectory data MT(t). If the system detects that the skipping rhythm has become unstable or the motion trajectory has abnormal fluctuations, it will reduce the value of T. For example, if the coefficient of variation of the interval time between 10 consecutive jumps exceeds 20%, the system will reduce the T value by 0.15.

[0191] The improvement potential (P) parameter adjusts relatively slowly because it reflects a long-term trend. The system adjusts this parameter based on the jumper's performance changes over a longer period of time (such as weeks or months). If the system observes that the jumper's various indicators are gradually improving, it will gradually increase the P value; conversely, if there is no obvious improvement or regression over a long period of time, the P value will gradually decrease.

[0192] These parameter adjustments are a dynamic and continuous process. The system adjusts parameters based on different time windows, including short windows (such as 0.5 seconds and 5 seconds) to capture instantaneous changes and short-term trends, and long windows (such as 30 seconds and the entire training process) to evaluate overall performance and long-term trends. This multi-time scale analysis ensures that the system can fully capture the various changes in the rope skipping process.

[0193] When making these adjustments, the system continuously considers the jumper's current condition category. For jumpers in the "excellent" category, the parameter adjustments are relatively small to avoid oversensitivity. For jumpers in the "potential risk" category, the same anomalies will result in larger parameter adjustments. This differentiated adjustment ensures that the system can provide appropriate safety assessments for jumpers of different levels.

[0194] Finally, based on these dynamically adjusted risk assessment parameters, the system calculates the comprehensive safety index (CSI). The calculation formula is:

[0195] CSI=100*(w1*S+w2*(1-F)+w3*T+w4*I+w5*P) / (w1+w2+w3+w4+w5)

[0196] Where w1 to w5 are the weights of each parameter, which directly correspond to the edge weights of the weighted directed graph in step 130. The calculated CSI is a score from 0 to 100, where 100 represents the safest and 0 represents the highest risk. The system maps this score back to five status categories to provide a more refined safety status assessment.

[0197] In this way, step S140 realizes the transition from the qualitative analysis of step S130 to the quantitative evaluation. This process can be illustrated by a specific example: suppose the system detects that the jumper's arm position drops due to fatigue at 24.62 seconds, triggering the analysis condition related to RP(t). This triggers a series of chain reactions:

[0198] 1. At 24.62 seconds, the arm position dropped, triggering the analysis condition related to RP(t). The system reduced the action safety (S) parameter from the initial value of 0.8 to 0.75.

[0199] 2. At 24.82 seconds, the system detected a compensatory forward lean of the trunk, triggering the analysis condition of P(t). This caused the S value to further decrease to 0.7, while the fatigue risk (F) value increased from the initial 0.3 to 0.4.

[0200] 3. At 24.95 seconds, the rope tension increases, triggering the T(t) analysis condition. The system reduces the strength adequacy (I) value from 0.75 to 0.65.

[0201] 4. Finally at 25 seconds, the system detected a significant speed drop, triggering the main analysis condition of V(t). This caused the skill stability (T) value to drop from 0.8 to 0.7.

[0202] Each triggered analysis step results in the adjustment of the corresponding risk assessment parameters. These adjustments combined lead to significant changes in the Comprehensive Safety Index (CSI). Assuming that the initial CSI is 85 points, which is in the "good" range, after this series of events, the CSI drops to 65 points, entering the "needs improvement" range.

[0203] Step S150, executing early warning and recommended measures based on the comprehensive safety indicators and in-depth analysis results.

[0204] In one example, the step of executing warnings and recommended measures based on the comprehensive safety indicators and in-depth analysis results includes the steps of: determining the rope jumper's actions as a safe level, a low risk level, a medium risk level, a high risk level, or an extremely high risk level based on the comprehensive safety indicators and in-depth analysis results;

[0205] Execute corresponding warnings and recommended measures based on the security level, low risk level, medium risk level, high risk level or extremely high risk level.

[0206] 1. Safety level (CSI ≥ 90): The rope skipping activity is in an ideal state, with standardized movements and moderate intensity. The system maintains normal monitoring and does not intervene specifically. It will regularly send positive feedback to the rope jumper, such as "Keep in good shape", to encourage continued correct rope skipping posture and rhythm. The system will send such positive feedback every 5 minutes to avoid excessively disturbing the rope jumper while maintaining moderate encouragement. If in-depth analysis finds that certain specific movements are particularly important for maintaining a high level of performance, the system may emphasize these movements in the feedback, such as "Your wrist angle is well controlled, which is essential to maintaining efficient rope skipping."

[0207] 2. Low risk level (80≤CSI<90): There are slight deviations in the rope skipping activity, but it is still within an acceptable range overall. The system uses visual cues to alert the rope jumper to potential problems. For example, a yellow warning sign is displayed on a display screen within the rope jumper's field of vision. At the same time, the system sends specific suggestions for improvement, such as "slightly increase the elbow position" or "keep the jumping rhythm more even." The system will also increase the monitoring frequency and reassess the risk level every 2 minutes to ensure that changes in risk can be captured in a timely manner. If in-depth analysis identifies the specific reasons for the slight deviation, the system may provide more precise suggestions, such as "Your left arm is slightly lower than your right arm by 2 cm. Try to raise your left arm slightly to improve your balance."

[0208] 3. Medium risk level (70≤CSI<80): There are obvious technical or physical problems in the rope skipping activity, which requires attention. The system will issue a clear sound warning to remind the rope jumper to pay attention to the current situation. At the same time, specific risk points are displayed in eye-catching orange font on the display screen, such as "knee bending angle is too large" or "jumping frequency is unstable". The system will also recommend that rope jumpers take a short break or slow down the pace, and provide detailed adjustment guidance. At this risk level, the system reassesses the risk level every minute to ensure that it can respond quickly to changes in risks. Combined with the results of in-depth analysis, the system may provide more comprehensive adjustment suggestions, such as "Excessive bending of your knees may be related to unstable rope skipping rhythm. Try to keep your knees slightly bent while listening to the metronome to maintain a stable rhythm, which helps reduce knee pressure and improve overall coordination."

[0209] 4. High risk level (60≤CSI<70): There are serious movement errors or excessive fatigue, which may lead to injury. The system issues a strong sound and light warning, requiring the jumper to pay attention immediately. The warning message and specific risks will be displayed in red flashing on the display. The system strongly recommends that the jumper stop the current action immediately, make adjustments or rest. If the jumper is wearing a smart device, the system will send a vibration warning through the device. At this risk level, the system reassesses the risk level every 30 seconds until the risk is reduced. If an in-depth analysis finds a specific chain of causes leading to high risk, the system will provide more detailed warnings and suggestions, such as "Warning: Postural problems caused by fatigue are detected that are increasing the risk of injury. It is recommended to stop immediately, rest for 5 minutes, and perform the following specific stretching movements...".

[0210] 5. Extremely high risk level (CSI <60): Rope skipping activity is in a dangerous state and there is an immediate risk of injury. The system triggers the emergency warning mechanism and issues the highest level of sound and light alarm. If the rope skipping equipment supports it, the system automatically increases the rope resistance or triggers the safety stop mechanism. The system sends an emergency stop command to the jumper and provides immediate first aid instructions. If the system is connected to a gym or coaching terminal, it will automatically notify relevant personnel to provide assistance. Based on the results of in-depth analysis, the system may predict potential acute injury risks and provide specific warnings, such as "Warning! The current landing method is very likely to cause Achilles tendon strain. Please stop skipping immediately and sit down to rest."

[0211] In this way, the system performs early warnings and recommended actions based primarily on comprehensive safety indicators, while combining in-depth analysis results where appropriate to provide more precise and targeted feedback. This approach maintains the timeliness of real-time feedback while using in-depth analysis results to improve the precision and personalization of recommendations when necessary.

[0212] In one application embodiment:

[0213] Xiao Ming uses the intelligent rope skipping safety monitoring system for training. The system first quickly collects, simplifies and normalizes various data (step S110), then extracts key features, calculates Xiao Ming's current rope skipping status score (step S110), and builds a data traceability relationship (step S120).

[0214] In step S130, the system designs a reverse tracking path. This path is like a timeline, marking the key data points of Xiao Ming's skipping process. The system detects that at 24.62 seconds, Xiao Ming's arm position drops, which is the first key data point. Subsequently, the system captures a series of key data points: the torso shows a compensatory forward lean at 24.82 seconds, the rope tension increases at 24.95 seconds, and finally the speed drops significantly at 25 seconds.

[0215] The system immediately initiated a deep analysis process, which quickly back-checked, expanded, cross-validated, and contextualized these key data points. The results showed a causal relationship between these data points, reflecting the impact of fatigue accumulation on Xiao Ming's rope skipping performance.

[0216] Based on the analysis results of these key data points, the system dynamically adjusts the risk assessment parameters in step S140 and calculates a comprehensive safety index (CSI) of 87 points, indicating that Xiao Ming is at a low risk level.

[0217] Finally, in step S150, based on the analysis results of CSI and key data points, the system reminds Xiao Ming to pay attention to posture changes and signs of fatigue through visual prompts, and provides targeted improvement suggestions, such as adjusting the arm position and paying attention to maintaining a correct torso posture.

[0218] The present application is carried out through the following steps: collecting video data of rope jumpers in rope skipping activities and preprocessing; extracting key features from the preprocessed data, and obtaining rope skipping status categories based on the key features; constructing a data traceability relationship network based on the key features; the data traceability relationship network is used to display the direct relationship and indirect impact path between key features; designing a data inspection mechanism based on the rope skipping status categories and the data traceability relationship network, and setting corresponding in-depth analysis conditions and trigger mechanisms for key parameters respectively, and implementing an in-depth analysis process to obtain in-depth analysis results when the in-depth analysis conditions and trigger mechanisms are achieved; defining risk assessment parameters, initializing risk assessment parameters based on rope skipping status categories, dynamically adjusting the initialized risk assessment parameters based on the data inspection mechanism and the trigger mechanism, and calculating comprehensive safety indicators based on the dynamically adjusted risk assessment parameters; executing early warning and recommended measures based on the comprehensive safety indicators and in-depth analysis results. The present application obtains the potential risks of rope jumpers in the rope skipping process by performing a series of analyses on the video data of rope jumpers, and provides early warnings and suggestions to rope jumpers, thereby realizing safe monitoring of rope skipping activities and effectively identifying and warning potential safety risks.

[0219] The data traceability relationship constructed in step S120 of the present application reveals the complex relationship between the parameters in the rope skipping action by calculating the cross-correlation function and time delay between the parameters. This method enables the system to accurately trace the development process of the problem. For example, when an abnormal speed V(t) is detected, the system can trace back and find that this is due to the posture change P(t) 0.18 seconds ago. This precise timing association enables the system to deeply understand the origin and evolution of the problem, providing a basis for more comprehensive safety analysis and targeted improvement suggestions.

[0220] The reverse tracking path designed in step S130 provides a new method for problem diagnosis. When the system detects an anomaly, it can not only identify the direct cause, but also trace back to the root cause of the problem. For example, if a sudden drop in speed is detected, the system will trace back along the timeline and find that this is due to the accumulation of small posture changes that began 10 seconds ago. This deep analysis capability allows the system to provide more targeted improvement suggestions, not only to solve surface problems, but also to prevent similar problems from happening again.

[0221] The dynamic adjustment mechanism of the risk assessment parameters in step S140 makes the system highly adaptable. The system not only adjusts the parameters according to the preset condition categories, but also continuously fine-tunes them based on real-time feedback. For example, if it is found that the change in the body shape of a certain rope jumper has a particularly significant impact on the speed, the system will dynamically increase the weight of the relevant parameters. This adaptive mechanism enables the system to provide personalized safety assessments for rope jumpers of different levels and different physical characteristics, greatly improving the universality and accuracy of the system.

[0222] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0223] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 2 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store video data, rope skipping data and other data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an intelligent rope skipping sports safety warning method based on behavior recognition is implemented.

[0224] Those skilled in the art will understand that Figure 2The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0225] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0226] Collecting video data of a rope jumper during rope jumping activities, and obtaining rope jumping data from the video data; the rope jumping data includes rope jumping action image data, rope jumping frequency data, rope jumper posture data, and rope tension data;

[0227] Collect and pre-process the video data of the rope jumper during the rope jumping activity; extract key features from the pre-processed data, and obtain the rope jumping status category based on the key features; the key features include body posture, rope jumping speed, rope jumping rhythm, relative positions of various body parts, motion trajectory and rope tension;

[0228] According to the key features, a data traceability relationship network is constructed; the data traceability relationship network is used to show the direct relationship and indirect impact path between the key features;

[0229] According to the rope skipping status category and data traceability relationship network, a data inspection mechanism is designed, and corresponding in-depth analysis conditions and trigger mechanisms are set for key parameters. When the in-depth analysis conditions and trigger mechanisms are met, the in-depth analysis process is implemented to obtain in-depth analysis results; key parameters are one or any combination of key features;

[0230] Define risk assessment parameters, initialize the risk assessment parameters based on the rope skipping status category, dynamically adjust the initialized risk assessment parameters based on the data inspection mechanism and the trigger mechanism, and calculate the comprehensive safety index based on the dynamically adjusted risk assessment parameters;

[0231] Implement early warning and recommended measures based on comprehensive safety indicators and in-depth analysis results.

[0232] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0233] Collect and pre-process the video data of the rope jumper during the rope jumping activity; extract key features from the pre-processed data, and obtain the rope jumping status category based on the key features; the key features include body posture, rope jumping speed, rope jumping rhythm, relative positions of various body parts, motion trajectory and rope tension;

[0234] According to the key features, a data traceability relationship network is constructed; the data traceability relationship network is used to show the direct relationship and indirect impact path between the key features;

[0235] According to the rope skipping status category and data traceability relationship network, a data inspection mechanism is designed, and corresponding in-depth analysis conditions and trigger mechanisms are set for key parameters. When the in-depth analysis conditions and trigger mechanisms are met, the in-depth analysis process is implemented to obtain in-depth analysis results; key parameters are one or any combination of key features;

[0236] Define risk assessment parameters, initialize the risk assessment parameters based on the rope skipping status category, dynamically adjust the initialized risk assessment parameters based on the data inspection mechanism and the trigger mechanism, and calculate the comprehensive safety index based on the dynamically adjusted risk assessment parameters;

[0237] Implement early warning and recommended measures based on comprehensive safety indicators and in-depth analysis results.

[0238] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0239] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0240] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent application shall be subject to the attached claims.

Claims

1. An intelligent rope skipping sports safety warning method based on behavior recognition, characterized in that: The following steps are involved: Collect and pre-process video data of rope jumpers during rope jumping activities; Extracting key features from the preprocessed data, and obtaining the rope skipping status category according to the key features; the key features include body posture, rope skipping speed, rope skipping rhythm, relative positions of various parts of the body, motion trajectory and rope tension; According to the key features, a data traceability relationship network is constructed; The data traceability relationship network is used to display the direct relationship and indirect impact path between the key features; According to the rope skipping status category and the data traceability relationship network, a data inspection mechanism is designed, and corresponding in-depth analysis conditions and trigger mechanisms are set for key parameters respectively. When the in-depth analysis conditions and trigger mechanisms are met, an in-depth analysis process is implemented to obtain in-depth analysis results; the key parameter is one or any combination of the key features; Defining risk assessment parameters, initializing the risk assessment parameters based on the rope skipping status category, dynamically adjusting the initialized risk assessment parameters based on the data checking mechanism and the triggering mechanism, and calculating a comprehensive safety index based on the dynamically adjusted risk assessment parameters; Implement early warning and recommended measures based on the comprehensive safety indicators and the in-depth analysis results; The step of collecting video data of a rope jumper during rope jumping and preprocessing the video data includes the following steps: Collecting the video data of the rope jumper; extracting multi-dimensional rope jumping data from the video data, wherein the multi-dimensional rope jumping data includes rope jumping action image data, rope jumping frequency data, rope jumping body posture data and rope tension data; Simplifying the multi-dimensional rope skipping data to obtain simplified data; Performing normalization processing on the simplified data to obtain normalized data; The step of extracting key features from the preprocessed data and obtaining the rope skipping status category according to the key features comprises the following steps: Using a dynamic time warping algorithm to compare the key feature with features corresponding to a preset condition category, and obtaining a similarity score corresponding to the key feature; Performing weighted processing on the similarity scores to obtain a comprehensive score; Based on the comprehensive score, identifying the rope skipping condition category; Among them, it is characterized in that, according to the key features, the step of constructing a data traceability relationship network includes the steps of: Calculating the cross-correlation function and conditional probability distribution between the key features; Based on multiple regression analysis, the dynamic relationship between the key features is obtained; constructing a weighted directed graph based on the cross-correlation function, the conditional probability distribution and the dynamic relationship; Calculating edge weights of the weighted directed graph based on the maximum value of the cross-correlation function, the strength of the conditional probability distribution, and the regression coefficient of the dynamic relationship; Based on the weighted directed graph and the edge weights, constructing the data traceability relationship network; According to the rope skipping status category and the data traceability relationship network, a data inspection mechanism is designed, and corresponding in-depth analysis conditions and trigger mechanisms are set for key parameters respectively. When the in-depth analysis conditions and trigger mechanisms are met, the in-depth analysis process is implemented to obtain the in-depth analysis results, including the following steps: Design a basic tracking strategy based on the skipping rope condition categories; After designing the basic tracking strategy, determining the key parameters to be tracked according to the weighted directed graph corresponding to the data traceability relationship network; After determining the key parameters, designing the data checking mechanism based on the time delay relationship of the weighted directed graph; After designing the data inspection mechanism, corresponding in-depth analysis conditions and trigger mechanisms are set for the key parameters respectively; When the in-depth analysis conditions and trigger mechanisms are met, rapid backtracking checks, expanded inspection scope, cross-validation, contextual analysis and dynamic weight adjustments are implemented to implement the in-depth analysis process and obtain in-depth analysis results.

2. The intelligent rope skipping sports safety early warning method based on behavior recognition according to claim 1 is characterized in that: The step of calculating the edge weight of the weighted directed graph based on the maximum value of the cross-correlation function, the strength of the conditional probability distribution and the regression coefficient of the dynamic relationship comprises: Determining key factors for calculating the edge weight, the key factors comprising the maximum value of the cross-correlation function, the strength of the conditional probability distribution, and the regression coefficient of the dynamic relationship; Setting weight coefficients of the key factors; Based on the key factors and the weight coefficients, a comprehensive calculation model of the edge weight is established; The comprehensive calculation model is used to calculate the weight of each edge in the weighted directed graph to obtain the edge weight of the weighted directed graph.

3. The intelligent rope skipping sports safety early warning method based on behavior recognition according to claim 1 is characterized in that: When the in-depth analysis conditions and trigger mechanism are met, the steps of implementing rapid backtracking inspection, expanding the inspection scope, cross-validation, context analysis and dynamic weight adjustment to achieve the in-depth analysis process and obtain in-depth analysis results include: Perform the quick backtracking check to analyze the changes of the key parameters triggering the in-depth analysis within a predetermined period of time in the past; When the rapid retrospective inspection does not reveal obvious abnormalities, the expanded inspection scope is performed to extend the analysis to a longer time period and more relevant parameters; Perform the cross-validation, check the changes of other key parameters, and analyze the mutual influence between the key parameters; The context analysis is performed by combining the historical data and the current training plan of the rope jumper; Based on the results of the fast backtracking check, the expanded check range, the cross validation and the context analysis, the weights are dynamically adjusted to update the edge weights in the weighted directed graph; The in-depth analysis result is obtained by integrating the results of the rapid backtracking check, the expanded check range, the cross-validation, the context analysis and the dynamic adjustment of the weights.

4. The intelligent rope skipping sports safety early warning method based on behavior recognition according to claim 1 is characterized in that: Defining risk assessment parameters, initializing the risk assessment parameters based on the rope skipping status category, dynamically adjusting the initialized risk assessment parameters based on the data checking mechanism and the triggering mechanism, and calculating a comprehensive safety index based on the dynamically adjusted risk assessment parameters; Defining the risk assessment parameters; the risk assessment parameters include action safety, fatigue risk, skill stability, intensity appropriateness, and improvement potential; Initializing the action safety, the fatigue risk, the skill stability, the intensity appropriateness, and the improvement potential based on the rope skipping condition category; Based on the data checking mechanism and the triggering mechanism, dynamically adjusting the action safety, the fatigue risk, the skill stability, the intensity appropriateness and the improvement potential for the initialized rope skipping condition category; The dynamically adjusted action safety, fatigue risk, skill stability, intensity appropriateness and improvement potential are weighted to obtain the comprehensive safety index.

5. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.