Image-based AI intelligent exercise health detection method and system

Abstract

The invention discloses an image-based AI intelligent exercise health detection method and system, and the method comprises the steps: extracting human body key point coordinate data and a facial feature region in a continuous video image; calculating a displacement change rate of the key point based on the human body key point coordinate data, and obtaining a movement speed parameter according to the displacement change rate; extracting color feature parameters of the facial feature region; performing time sequence correlation on the motion speed parameters and the color characteristic parameters of the corresponding time points to obtain a motion physiological parameter correlation curve; segmenting the exercise physiological parameter correlation curve, and calculating a parameter fluctuation range of each segment of the exercise physiological parameter correlation curve; determining a change trend between the parameter fluctuation ranges of the adjacent preset time windows, and generating a trend index based on the change trend; and comparing the trend index with a preset reference interval to obtain a comparison result, and generating an exercise health detection report based on the comparison result. The accuracy of monitoring the exercise health state of the user is improved.

Description

Technical Field

[0001] This application belongs to the field of sports health detection, and particularly relates to an AI intelligent sports health detection method and system based on images. Background Art

[0002] With the increasing emphasis on a healthy lifestyle by people, sports health detection has gradually become an important research field. Traditional sports health detection mainly relies on wearable devices and professional medical devices. These devices are not only expensive, but also require users to actively wear them or go to medical institutions for detection. The usage process is cumbersome and it is difficult to achieve real-time monitoring in daily life, resulting in poor continuity and integrity of detection data and being unable to provide users with timely and effective health warnings and sports guidance.

[0003] In related technologies, image data of a user during exercise can be collected through a camera, and deep learning algorithms can be used to identify and track human key points to achieve real-time analysis of exercise postures. This technology does not require users to wear any devices and can be detected in a natural state, improving the detection efficiency.

[0004] However, the above-mentioned image-based exercise posture recognition technology mainly focuses on static posture analysis at a single time point. When the exercise intensity is relatively high, it is difficult to accurately judge the degree of physical fatigue of users only relying on posture data, reducing the accuracy of monitoring sports health status. Summary of the Invention

[0005] This application provides an AI intelligent sports health detection method and system based on images, which is used to improve the accuracy of monitoring the sports health status of users.

[0006] In a first aspect, this application provides an AI intelligent sports health detection method based on images, which extracts the coordinate data of human key points and facial feature regions in continuous video images of a user collected by a camera device; Calculates the displacement change rate of the key points based on the coordinate data of the human key points, and obtains a motion speed parameter according to the displacement change rate; Performs skin color analysis on the facial feature region, extracts color feature parameters, and the color feature parameters include the ratio of the red component of the user's facial region; Performs time series association on the motion speed parameter and the color feature parameter at the corresponding time point to obtain a motion physiological parameter association curve; Segments the motion physiological parameter association curve according to a preset time window, and calculates the parameter fluctuation range of each segment of the motion physiological parameter association curve; Determines the change trend between the parameter fluctuation ranges of adjacent preset time windows, and generates a trend index based on the change trend; Compare the trend index with a preset reference interval to obtain a comparison result, and generate a sports health detection report based on the comparison result.

[0007] By adopting the above technical solution, by extracting the human key point coordinate data and facial feature regions in consecutive video images, performing temporal correlation in combination with the motion speed parameter and color feature parameter, and generating a motion physiological parameter correlation curve, real-time monitoring of the motion state can be achieved. Performing segmented processing on the correlation curve and calculating the parameter fluctuation range can reflect the changes in the motion intensity within different time windows. By calculating the change trend of the parameter fluctuation range in adjacent time windows, generating a trend index and comparing it with the preset reference interval, the accuracy of evaluating the physiological load level during the motion process can be improved, and the convenience of detection is enhanced. By fusing the data in two dimensions of motion speed and facial skin color change, the reliability and accuracy of the detection result are enhanced.

[0008] Combined with some embodiments of the first aspect, in some embodiments, calculate the displacement change rate of the key points based on the human key point coordinate data, and obtain the motion speed parameter according to the displacement change rate. Specifically, it includes: Establish a relative motion coordinate system with the center of the user's torso as the origin, and map the human key point coordinate data to the relative motion coordinate system; Calculate the three-dimensional displacement vector of the corresponding key points in consecutive video images of adjacent frames, and obtain the displacement change rate of each key point according to the position change of each key point in the relative motion coordinate system; Convert the displacement change rate of each key point into a motion speed parameter according to the acquisition frame rate of the camera device.

[0009] By adopting the above technical solution, by establishing a relative motion coordinate system with the center of the user's torso as the origin, the interference caused by the change of the camera device position or the overall displacement of the user can be reduced, ensuring that the obtained motion parameters only reflect the relative motion state of each human key point. After mapping the human key point coordinate data to the relative motion coordinate system, calculating the three-dimensional displacement vector between adjacent frames of key points can accurately capture the spatial characteristics of human motion. Combining the acquisition frame rate of the camera device to convert the displacement change rate into a motion speed parameter can obtain standardized speed data, improving the accuracy and reliability of the speed parameter, enabling the system to accurately identify and quantify different types of motion patterns.

[0010] Combined with some embodiments of the first aspect, in some embodiments, determine the change trend between the parameter fluctuation ranges of adjacent preset time windows, and generate a trend index based on the change trend. Specifically, it includes: Calculate the difference between the parameter fluctuation ranges of the motion physiological parameter correlation curves of adjacent preset time windows to obtain a fluctuation range difference value; Divide the difference value of the fluctuation range by the time interval of adjacent preset time windows to obtain the change rate of the parameter fluctuation range; Determine the change trend of the correlation curve of the exercise physiological parameter according to the positive or negative of the change rate. A positive change rate indicates an upward trend, and a negative change rate indicates a downward trend; Perform weighted calculation on the change rate of the parameter fluctuation range based on the change trend to obtain the weighted change rate; Map the weighted change rate into a preset interval to generate a trend index.

[0011] By adopting the above technical solution, by calculating the difference value of the fluctuation range of the correlation curve of the exercise physiological parameter in adjacent preset time windows and combining the time interval to obtain the change rate of the parameter fluctuation range, the dynamic change process of the exercise intensity can be quantitatively reflected. Judging the change trend of the correlation curve of the exercise physiological parameter according to the positive or negative of the change rate and performing weighted calculation to obtain the weighted change rate can highlight important physiological change characteristics. The processing method of mapping the weighted change rate into a preset interval to generate a trend index enables the physiological changes in different exercise states to be compared and evaluated in a standardized manner, improving the accuracy and reliability of exercise health detection.

[0012] Combined with some embodiments of the first aspect, in some embodiments, after comparing the trend index with a preset reference interval to obtain a comparison result and generating an exercise health detection report based on the comparison result, the method further includes: Determine the distribution of the peaks and valleys of the correlation curve of the exercise physiological parameter; Determine the exercise cycle characteristics of the user according to the distribution of the peaks and valleys; Segment the trend index based on the exercise cycle characteristics and perform respiratory rhythm analysis on each segment of the trend index to obtain the respiratory frequency and respiratory depth; Fit the respiratory frequency and respiratory depth with the trend index to obtain a matching coefficient; When the matching coefficient is greater than a preset synchronization threshold, increase the upper limit value of the preset reference interval by a preset first percentage and decrease the lower limit value by a preset second percentage; When the matching coefficient is less than a preset synchronization threshold, decrease the upper limit value of the preset reference interval by a preset first percentage and increase the lower limit value by a preset second percentage.

[0013] By adopting the above technical solution, the characteristics of the movement cycle are determined by analyzing the distribution of the peaks and valleys of the correlation curve of the exercise physiological parameters, and based on this, the trend index is segmented for respiratory rhythm analysis, so as to conduct a correlation analysis between the exercise intensity and the respiratory state. By calculating the matching coefficients of the respiratory frequency, respiratory depth and trend index, the upper and lower limit values of the preset reference interval are dynamically adjusted, enabling the system to adaptively adjust the evaluation criteria according to the user's respiratory state. When the respiratory rhythm is highly matched with the exercise intensity, the reference interval is expanded, and vice versa, the reference interval is narrowed, improving the personalization and accuracy of the exercise health detection, and making the detection results more in line with the user's actual physiological state.

[0014] Combined with some embodiments of the first aspect, in some embodiments, the respiratory frequency and respiratory depth are fitted with the trend index to obtain a matching coefficient, specifically including: Denote the time series curve of the respiratory frequency as the first curve, the time series curve of the respiratory depth as the second curve, and the time series curve of the trend index as the third curve; Calculate the Pearson correlation coefficient between the first curve and the third curve and the Pearson correlation coefficient between the second curve and the third curve respectively; Normalize the first curve and the second curve to obtain a normalized curve; Synthesize the normalized curves into a comprehensive curve by weighted summation; Calculate the Euclidean distance between the comprehensive curve and the third curve, and take the reciprocal of the Euclidean distance as the matching coefficient.

[0015] By adopting the above technical solution, through the correlation analysis and comprehensive fitting of the respiratory frequency time series curve, respiratory depth time series curve and trend index time series curve, the coordination of physiological parameters during exercise can be evaluated from multiple dimensions. Calculating the Pearson correlation coefficient can reflect the linear correlation degree between the respiratory frequency, respiratory depth and exercise trend. The comprehensive curve obtained by weighted summation contains the comprehensive characteristics of the respiratory frequency and respiratory depth, and calculating the Euclidean distance with the trend index curve can quantitatively evaluate the matching degree between the respiratory rhythm and the exercise trend. Taking the reciprocal of the Euclidean distance as the matching coefficient can make the higher the matching degree, the larger the coefficient value, which is convenient for dynamically adjusting the range of the preset reference interval according to the matching coefficient later, improving the accuracy and scientificity of the exercise health status evaluation.

[0016] Combined with some embodiments of the first aspect, in some embodiments, after reducing the upper limit value of the preset reference interval by a preset first percentage and increasing the lower limit value by a preset second percentage, the method further includes: Obtain the peak points and valley points of adjacent multiple movement cycles; Calculate the peak slope between adjacent peak points and the valley slope between adjacent valley points; Determine the fatigue compensation coefficient according to the ratio of the peak slope to the valley slope; Multiply the matching coefficient by the fatigue compensation coefficient to obtain the corrected matching coefficient.

[0017] By adopting the above technical solution, by analyzing the change characteristics of the peak points and valley points of adjacent motion cycles, calculating the ratio of the peak slope to the valley slope to determine the fatigue compensation coefficient, the fatigue state during the motion can be effectively identified. The peak slope reflects the change trend of the exercise intensity, and the valley slope reflects the change trend of the recovery ability. The ratio of the two can characterize the dynamic balance relationship between exercise and recovery. Multiplying the matching coefficient by the fatigue compensation coefficient to obtain the corrected matching coefficient introduces the influence of the fatigue factor on the basis of the original matching evaluation of the breathing rhythm and the motion trend, making the evaluation of the exercise health state more comprehensive and accurate. The corrected matching coefficient can better reflect the exercise adaptation ability of the user under different fatigue degrees, and helps to detect potential exercise risks in a timely manner.

[0018] Combined with some embodiments of the first aspect, in some embodiments, determining the fatigue compensation coefficient according to the ratio of the peak slope to the valley slope specifically includes: When the peak point slope is positive and the valley point slope is negative, set the fatigue compensation coefficient to the first preset value; When the peak point slope is negative and the valley point slope is negative, set the fatigue compensation coefficient to the second preset value; When the peak point slope is negative and the valley point slope is positive, set the fatigue compensation coefficient to the third preset value, the first preset value is greater than the second preset value, and the second preset value is greater than the third preset value.

[0019] By adopting the above technical solution, by classifying and judging the positive and negative of the peak slope and the valley slope, and setting different sizes of fatigue compensation coefficients according to different combinations, a reasonable fatigue degree quantification method is established. When the peak slope is positive and the valley slope is negative, it indicates that the exercise intensity is increasing while the recovery ability is decreasing. At this time, the largest fatigue compensation coefficient is set, reflecting a higher exercise risk; when both the peak slope and the valley slope are negative, it indicates that both the exercise intensity and the recovery ability are decreasing, and a medium-sized fatigue compensation coefficient is set; when the peak slope is negative and the valley slope is positive, it indicates that the exercise intensity is decreasing while the recovery ability is increasing. At this time, the smallest fatigue compensation coefficient is set, reflecting a lower exercise risk, improving the accuracy of the system to identify the exercise risks in different fatigue states.

[0020] In a second aspect, an embodiment of the present application provides an AI intelligent motion health detection system based on images. The AI intelligent motion health detection system based on images includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code. The computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the system to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, including instructions, which when running on the system, cause the system to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer program product, characterized in that when the computer program product runs on the system, it causes the system to execute the method described in any possible implementation manner in the first aspect.

[0023] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. The present application provides an AI intelligent motion health detection method based on images. By extracting the human key point coordinate data and facial feature regions in continuous video images, and performing time series association by combining the motion speed parameter and the color feature parameter, a motion physiological parameter association curve is generated, which can realize the real-time monitoring of the motion state. By segmenting the association curve and calculating the parameter fluctuation range, the change of the motion intensity within different time windows can be reflected. By calculating the change trend of the parameter fluctuation range in adjacent time windows, generating a trend index and comparing it with a preset reference interval, the accuracy of evaluating the physiological load level during the motion process can be improved, and the convenience of detection is improved. By fusing the data of two dimensions, namely the motion speed and the change of facial skin color, the reliability and accuracy of the detection result are enhanced.

[0024] 2. The present application provides an AI intelligent motion health detection method based on images. By analyzing the distribution of the peaks and valleys of the motion physiological parameter association curve to determine the motion cycle characteristics, and segmenting the trend index based on this for respiratory rhythm analysis, the motion intensity and the respiratory state can be associated and analyzed. By calculating the matching coefficients of the respiratory frequency, respiratory depth and the trend index, the upper and lower limits of the preset reference interval are dynamically adjusted, so that the system can adaptively adjust the evaluation criteria according to the user's respiratory state. When the respiratory rhythm is highly matched with the motion intensity, the reference interval is expanded, otherwise it is narrowed, which improves the personalization degree and accuracy of the motion health detection, and makes the detection result more in line with the actual physiological state of the user.

[0025] 3. This application provides an AI intelligent motion health detection method based on images. By analyzing the change characteristics of peak points and valley points in adjacent motion cycles and calculating the ratio of the peak slope to the valley slope to determine the fatigue compensation coefficient, the fatigue state during exercise can be effectively identified. The peak slope reflects the change trend of exercise intensity, and the valley slope reflects the change trend of recovery ability. The ratio of the two can characterize the dynamic balance relationship between exercise and recovery. Multiplying the matching coefficient by the fatigue compensation coefficient gives the corrected matching coefficient, introducing the influence of the fatigue factor on the basis of the original matching evaluation of respiratory rhythm and motion trend, making the evaluation of motion health status more comprehensive and accurate. The corrected matching coefficient can better reflect the exercise adaptation ability of users under different fatigue levels, helping to timely detect potential exercise risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 FIG. is a schematic flowchart of an AI intelligent motion health detection method based on images in an embodiment of this application.

[0027] Figure 2 FIG. is a schematic flowchart of a dynamic adjustment method for a preset reference interval in an embodiment of this application.

[0028] Figure 3 FIG. is a schematic structural diagram of an entity device of an AI intelligent motion health detection system provided in an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] The terms used in the following embodiments of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. As used in the specification and appended claims of this application, the singular forms "a", "an", "the", "above-mentioned", "the foregoing", "this" are also intended to include the plural forms, unless clearly indicated to the contrary in the context. It should also be understood that the term "and / or" used in this application refers to any or all possible combinations including one or more of the listed items.

[0030] Hereinafter, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as implying or indicating relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the embodiments of this application, unless otherwise stated, the meaning of "a plurality" is two or more.

[0031] The following uses an embodiment and in combination with Figure 1 to describe an AI intelligent motion health detection method based on images in an embodiment of this application: Please refer toFigure 1 This is a schematic flowchart of a method for AI intelligent motion health detection based on images in an embodiment of the present application.

[0032] S101. Extract the coordinate data of human key points and the facial feature region in the continuous video images collected by the camera device during the user's movement process; In this step, the system collects continuous video images of the user's movement process through the camera device, and extracts the coordinate data of human key points and the facial feature region from the video images. The coordinate data of human key points refers to the spatial coordinate position information of the key parts of the human body (such as the head, shoulders, elbows, wrists, hips, knees, ankles, etc.) in the video images. The facial feature region refers to the image region containing the user's face, which can be used for subsequent facial skin color analysis.

[0033] To implement this step, the system can use human pose estimation algorithms (such as OpenPose, AlphaPose, etc.) to process the video images, automatically detect and track human key points, and obtain their coordinate data. At the same time, the system can use face detection algorithms (such as Haar features, HOG features, etc.) to locate the facial region and extract it. In addition, the system can also combine deep learning methods, such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), etc., to further improve the detection accuracy of human key points and facial regions.

[0034] In practical applications, due to problems such as occlusion and blurring of the user's pose and face during the movement process, it is difficult to accurately extract the coordinate data of human key points and the facial feature region. To solve this problem, the system can adopt a multi-view video image acquisition method, simultaneously shoot the user's movement process from different angles through multiple camera devices, and then fuse the image information from different perspectives to obtain more complete and accurate coordinate data of human key points and facial feature regions. The fusion of multi-view video images can be achieved through technologies such as 3D reconstruction and perspective transformation.

[0035] S102. Calculate the displacement change rate of the key points based on the coordinate data of the human key points, and obtain the motion speed parameter according to the displacement change rate; The system calculates the displacement change rate of the key points based on the coordinate data of the human key points, and obtains the motion speed parameter according to the displacement change rate, specifically including: Establish a relative motion coordinate system with the center of the user's torso as the origin, and map the coordinate data of the human key points to the relative motion coordinate system; Calculate the three-dimensional displacement vectors of the corresponding key points in adjacent frames of continuous video images, and obtain the displacement change rate of each key point according to the position change of each key point in the relative motion coordinate system; Convert the displacement change rate of each key point into a motion speed parameter according to the acquisition frame rate of the imaging device.

[0036] In this step, the system calculates the displacement change rate of each key point in consecutive video images based on the obtained human key point coordinate data, and obtains the motion speed parameter according to the displacement change rate. The displacement change rate reflects the magnitude of the position change of each key point per unit time and can be used to measure the motion speed of the key point. The motion speed parameter is a quantitative index calculated based on the displacement change rate and is used to describe the overall motion speed level of the user.

[0037] To implement this step, the system first needs to establish a relative motion coordinate system with the center of the user's torso as the origin and map the human key point coordinate data into this coordinate system. Then, the system calculates the three-dimensional displacement vector of the corresponding key points in adjacent frame video images and obtains the displacement change rate of each key point according to the magnitude and direction of the displacement vector. Finally, the system considers the acquisition frame rate of the imaging device and converts the displacement change rate into a motion speed parameter with physical meaning, such as meters per second, etc.

[0038] In practical applications, since the displacement change of human key points may be affected by factors such as video image resolution and frame rate, the calculation accuracy of the motion speed parameter is reduced. To solve this problem, the system can adopt a multi-scale motion speed estimation method, analyze the displacement change of key points at different time and space scales, and obtain more robust and accurate motion speed parameters. In addition, the system can also combine a human kinematics model to physically constrain and optimize the displacement change of key points, further improving the reliability of the motion speed parameter.

[0039] S103. Perform skin color analysis on the facial feature area and extract color feature parameters; The system performs skin color analysis on the facial feature area and extracts color feature parameters. The color feature parameters include the ratio of the red component in the user's facial area.

[0040] In this step, the system performs skin color analysis on the extracted facial feature area and extracts color feature parameters that reflect the skin color change of the face. The color feature parameters can include the ratio of the red component in the facial area, etc., and are used to quantitatively describe the skin color change of the face. The skin color change of the face is closely related to the physiological state during human movement, such as blushing during strenuous exercise. Therefore, the color feature parameter can be used as an important index to evaluate the motion health state.

[0041] To implement this step, the system can convert the facial feature region into a specific color space (such as HSV, YCrCb, etc.), and then perform statistical analysis on different channels in the color space to obtain color feature parameters reflecting the facial skin color. For example, the system can calculate the proportion of the pixel values in the red channel in the facial region to the total pixel values to obtain the red component ratio. In addition, the system can also adopt machine learning methods, such as support vector machine (SVM), decision tree, etc., to automatically identify and extract the color feature parameters of the facial region by training a skin color classification model.

[0042] In practical applications, the skin color analysis of the facial feature region may be affected by factors such as lighting conditions and individual differences, resulting in inaccurate extraction of color feature parameters. To solve this problem, the system can adopt an adaptive skin color segmentation algorithm, which can adapt to different lighting conditions and individual skin color characteristics by dynamically adjusting the skin color segmentation threshold, thereby improving the robustness of skin color analysis. At the same time, the system can also introduce multiple color feature parameters, such as color histograms, color moments, etc., to comprehensively describe the variation characteristics of facial skin color and reduce the uncertainty brought by a single feature.

[0043] S104. Perform temporal correlation on the motion speed parameter and the color feature parameter at the corresponding time point to obtain a motion physiological parameter correlation curve. In this step, the system performs temporal correlation on the motion speed parameter and the color feature parameter at the corresponding time point to obtain a motion physiological parameter correlation curve reflecting the changes in physiological state during the motion process. The motion physiological parameter correlation curve reflects the dynamic relationship between the motion speed and the change in facial skin color, and can be used to analyze the physiological response characteristics of the user under different exercise intensities.

[0044] To implement this step, the system needs to pair the motion speed parameter and the color feature parameter in chronological order to form temporal data pairs. Then, the system can use techniques such as data interpolation and smoothing to process the temporal data to obtain a continuous motion physiological parameter correlation curve. The motion physiological parameter correlation curve can adopt various mathematical representation methods, such as function curves, parametric equations, etc., for subsequent analysis and modeling.

[0045] S105. Segment the motion physiological parameter correlation curve according to a preset time window, and calculate the parameter fluctuation range of each segment of the motion physiological parameter correlation curve. In this step, the system segments the motion physiological parameter correlation curve according to the preset time window, dividing the curve into multiple time segments. Then, the system calculates the parameter fluctuation range of the motion physiological parameter correlation curve within each time segment, that is, the difference between the maximum value and the minimum value of the parameter value. The parameter fluctuation range reflects the change amplitude of the motion physiological state within a certain time range and can be used to evaluate the stability of the physiological state during the motion process.

[0046] To implement this step, the system first needs to set an appropriate time window size according to the actual application requirements, such as 10 seconds, 30 seconds, etc. Then, the system equally divides the motion physiological parameter correlation curve according to the time window to obtain multiple time segments. For each time segment, the system calculates the maximum and minimum values of all parameter values on the curve within the segment, and the difference between the two is the parameter fluctuation range of the segment.

[0047] S106. Determine the change trend between the parameter fluctuation ranges of adjacent preset time windows, and generate a trend index based on the change trend; In this step, the system determines the change trend of the motion physiological state within a continuous time period by comparing the parameter fluctuation ranges of adjacent time windows. Then, the system generates a trend index based on the change trend to quantitatively describe the change amplitude and rate of the motion physiological state. The trend index can be used as an important indicator to evaluate the change of physiological state during exercise and is used to predict and warn of potential health risks.

[0048] To implement this step, the system first calculates the difference between the parameter fluctuation ranges of adjacent time windows to obtain a difference value reflecting the change amplitude of the fluctuation range. Then, the system divides the difference value by the time interval between adjacent time windows to obtain the change rate of the parameter fluctuation range, which reflects the change speed of the fluctuation range. According to the positive or negative sign of the change rate, the system can judge the change trend of the motion physiological state. A positive value indicates an upward trend, and a negative value indicates a downward trend. Finally, the system performs a weighted calculation on the change rate of the parameter fluctuation range to obtain a normalized trend index and maps it to a preset numerical interval for subsequent analysis and comparison.

[0049] In practical applications, the change trend of the parameter fluctuation range may be affected by factors such as individual differences and exercise intensity, resulting in deviations in the calculation results of the trend index. To solve this problem, the system can adopt an adaptive weight algorithm to dynamically adjust the weight coefficient of the change rate according to the physiological characteristics and exercise habits of different users, improving the individual adaptability of the trend index calculation. In addition, the system can also introduce a multi-index fusion strategy, comprehensively consider the change trends of multiple physiological parameters, and obtain a more comprehensive and reliable trend index through weighted combination.

[0050] S107. Compare the trend index with a preset reference interval to obtain a comparison result, and generate a motion health detection report based on the comparison result.

[0051] The system determines the change trend between the parameter fluctuation ranges of adjacent preset time windows and generates a trend index based on the change trend, specifically including: Calculate the difference between the parameter fluctuation ranges of the motion physiological parameter correlation curves of adjacent preset time windows to obtain a fluctuation range difference value; Divide the difference value of the fluctuation range by the time interval of adjacent preset time windows to obtain the change rate of the parameter fluctuation range; Determine the change trend of the motion physiological parameter correlation curve according to the positive or negative of the change rate. A positive change rate indicates an upward trend, and a negative change rate indicates a downward trend; Perform weighted calculation on the change rate of the parameter fluctuation range based on the change trend to obtain the weighted change rate; Map the weighted change rate to a preset interval to generate a trend index.

[0052] In this step, the system compares the calculated trend index with a preset reference interval to obtain a comparison result reflecting the change of physiological state during exercise. The reference interval is the normal range statistically obtained from the exercise physiological data of a large number of people, indicating the change range of the physiological state of healthy people during exercise. By comparing the trend index with the reference interval, the system can judge whether the user's exercise physiological state is at a normal level and timely detect potential health risks. Finally, the system generates an exercise health detection report based on the comparison result, providing intuitive and readable health assessment information and suggestions for the user.

[0053] To implement this step, the system needs to pre - construct a reference interval database covering different populations and exercise types. The reference interval can be obtained by statistical analysis of the exercise physiological data of a large - scale population or set according to the experience and suggestions of medical experts. When comparing, the system compares the trend index with the upper and lower boundary values of the reference interval to judge whether the trend index falls within the normal range. If the trend index exceeds the reference interval, it indicates that there may be abnormal risks in the user's exercise physiological state, and corresponding warning prompts need to be given. When generating the detection report, the system can use various forms such as text and charts to clearly display the comparison result and health suggestions, facilitating the user's understanding and implementation.

[0054] In practical applications, due to individual differences and the influence of exercise intensity, the setting of the reference interval may not be fully applicable to all users, resulting in deviations in the comparison results. To solve this problem, the system can adopt a personalized reference interval algorithm to dynamically adjust the boundary values of the reference interval according to factors such as the user's age, gender, and physical condition, improving the individual adaptability of the comparison results. In addition, the system can also introduce fuzzy evaluation technology. By establishing a fuzzy rule base, it makes a flexible judgment and description of the comparison results, avoiding the risk of misjudgment caused by hard thresholds.

[0055] In the above embodiments, by extracting the human key point coordinate data and facial feature regions in consecutive video images, and performing temporal correlation in combination with the motion speed parameter and color feature parameter to generate a motion physiological parameter correlation curve, real-time monitoring of the motion state can be achieved. By performing segmented processing on the correlation curve and calculating the parameter fluctuation range, the change of the motion intensity within different time windows can be reflected. By calculating the change trend of the parameter fluctuation range in adjacent time windows, generating a trend index and comparing it with a preset reference interval, the accuracy of evaluating the physiological load level during the motion process can be improved, and the convenience of detection is enhanced. By fusing the data in two dimensions of motion speed and facial skin color change, the reliability and accuracy of the detection result are enhanced.

[0056] After completing the above-mentioned trend index analysis of the motion physiological parameter correlation curve and generating the motion health detection report, the system can further dynamically adjust the preset reference interval in combination with the user's breathing rhythm characteristics to provide a more accurate assessment of the motion health state. The following combines Figure 2 to describe a method for dynamically adjusting a preset reference interval in an embodiment of the present application: Please refer to Figure 2 which is a schematic flowchart of a method for dynamically adjusting a preset reference interval in an embodiment of the present application.

[0057] S201. Determine the distribution of peaks and valleys of the motion physiological parameter correlation curve; In this step, the system analyzes the motion physiological parameter correlation curve to determine the distribution of peaks and valleys on the curve. The peak corresponds to the local maximum value of the physiological parameter during the motion process, reflecting the peak period of the physiological state; the valley corresponds to the local minimum value of the physiological parameter, reflecting the trough period of the physiological state. By analyzing the distribution characteristics of the peaks and valleys, the periodic change law of the physiological state during the motion process can be characterized.

[0058] To implement this step, the system can adopt extreme value detection algorithms in signal processing, such as the difference method, window scanning method, etc., to automatically identify the local maximum and minimum points on the motion physiological parameter correlation curve and obtain the position distribution of the peaks and valleys. In addition, the system can combine time-frequency analysis methods such as wavelet transform and empirical mode decomposition to further reveal the distribution law of the peaks and valleys of the motion physiological parameter correlation curve at different time scales.

[0059] In practical applications, since physiological signals during exercise are often affected by noise and interference, the accuracy of detecting peaks and valleys decreases. To solve this problem, the system can adopt an adaptive threshold algorithm, dynamically adjust the detection thresholds of peaks and valleys according to the local statistical characteristics of the correlation curve of exercise physiological parameters, and improve the robustness of detection. At the same time, the system can also introduce signal enhancement techniques such as morphological filtering and Kalman filtering to preprocess the correlation curve of exercise physiological parameters, suppress noise interference, and highlight the characteristics of peaks and valleys.

[0060] S202. Determine the exercise cycle characteristics of the user according to the distribution of peaks and valleys; In this step, the system determines the cycle characteristics of the user during exercise according to the distribution of peaks and valleys in the correlation curve of exercise physiological parameters. The exercise cycle characteristics reflect the rhythm and regularity of the change of the user's physiological state, and can be used to describe the user's exercise habits and physiological rhythm. By analyzing the exercise cycle characteristics, the system can master the law of the change of the user's physiological state during exercise, providing an important basis for subsequent trend index analysis and health assessment.

[0061] To implement this step, the system can calculate the time interval between adjacent peaks or valleys to obtain a time series reflecting the length of the exercise cycle. Then, the system conducts statistical analysis on the time series, such as calculating the mean value, variance, autocorrelation coefficient, etc. In addition, the system can also apply frequency domain analysis methods, such as Fourier transform and power spectrum analysis, to extract the distribution characteristics of the exercise cycle characteristics in the frequency domain, and reveal the dominant frequency and rhythm characteristics of the change of the user's physiological state.

[0062] In practical applications, the exercise cycle of the user may be affected by factors such as exercise intensity and environmental conditions, showing a certain degree of instability and variability. To solve this problem, the system can adopt an adaptive cycle tracking algorithm, adapt to the change of the user's exercise state by dynamically updating the statistical model of the exercise cycle, and improve the accuracy of cycle characteristic extraction. At the same time, the system can also introduce a multi-scale analysis strategy to extract exercise cycle characteristics on different time scales and comprehensively describe the multi-scale law of the change of the user's physiological state.

[0063] S203. Segment the trend index based on the exercise cycle characteristics, and conduct respiratory rhythm analysis on each segment of the trend index to obtain the respiratory frequency and respiratory depth; In this step, the system segments the trend index sequence based on the user's movement cycle characteristics. The purpose of segmentation is to match the trend index with the movement cycle, facilitating subsequent respiratory rhythm analysis. Then, the system performs respiratory rhythm analysis on the trend index segments within each movement cycle, extracting the respiratory frequency and respiratory depth characteristics that reflect the respiratory state. Respiratory frequency represents the number of breaths per unit time, and respiratory depth represents the amplitude of a single breath. By analyzing the respiratory rhythm characteristics, the system can evaluate the user's respiratory state during different movement cycles, providing an important reference for exercise health detection.

[0064] To implement this step, the system first divides the trend index sequence into segments corresponding to the movement cycle according to the movement cycle characteristics. Then, respiratory rhythm analysis is performed on each segment. The system can adopt an amplitude detection-based method to identify the peaks and valleys of the respiratory waveform, calculate the time interval between adjacent peaks as the respiratory frequency, and calculate the amplitude difference between the peak and the valley as the respiratory depth. In addition, the system can also apply time-frequency analysis techniques, such as short-time Fourier transform, wavelet transform, etc., to extract the energy distribution characteristics of the trend index segment in the respiratory frequency band, further characterizing the time-frequency domain characteristics of the respiratory rhythm.

[0065] In practical applications, since the respiratory signal is often coupled with other physiological signals, the accuracy of respiratory rhythm analysis decreases. To solve this problem, the system can adopt an adaptive signal separation algorithm. By dynamically estimating the coupling relationship between the respiratory signal and other physiological signals, the system can achieve the adaptive extraction and enhancement of the respiratory signal, improving the reliability of respiratory rhythm feature analysis. At the same time, the system can also introduce a multi-sensor fusion strategy, comprehensively utilizing the information of multiple physiological sensors (such as electrocardiogram, blood oxygen, etc.). Through signal fusion and complementarity, the stability and accuracy of respiratory rhythm feature extraction are improved.

[0066] S204. Fit the respiratory frequency and respiratory depth with the trend index to obtain a matching coefficient; The system fits the respiratory frequency and respiratory depth with the trend index to obtain a matching coefficient, which specifically includes: Denote the time series curve of the respiratory frequency as the first curve, the time series curve of the respiratory depth as the second curve, and the time series curve of the trend index as the third curve; Calculate the Pearson correlation coefficient between the first curve and the third curve and the Pearson correlation coefficient between the second curve and the third curve respectively; Normalize the first curve and the second curve to obtain a normalized curve; Synthesize the normalized curves into a comprehensive curve by weighted summation; Calculate the Euclidean distance between the comprehensive curve and the third curve, and take the reciprocal of the Euclidean distance as the matching coefficient.

[0067] In this step, the system fits the extracted respiratory rate and respiratory depth features with the trend index to obtain a matching coefficient reflecting the correlation between the two. The matching coefficient quantifies the degree of synchronization between the respiratory rhythm features and the changing trend of the exercise physiological state, and can be used as an important indicator to evaluate the association strength between the respiratory rhythm and the exercise health state. The higher the matching coefficient, the better the consistency between the respiratory rhythm and the changing trend of the exercise physiological state, and the stronger the indication of the respiratory rhythm on the exercise health state.

[0068] To implement this step, the system first aligns the time series of the respiratory rate, respiratory depth, and trend index to ensure their corresponding relationships on the time axis. Then, the system calculates the Pearson correlation coefficients between the respiratory rate and respiratory depth and the trend index respectively to measure the linear correlation between the two time series. Next, the system normalizes the respiratory rate and respiratory depth to eliminate the dimensional differences, and synthesizes them into a comprehensive respiratory rhythm feature sequence by weighted summation. Finally, the system calculates the Euclidean distance between the comprehensive respiratory rhythm feature sequence and the trend index sequence, and takes the reciprocal of the distance as the final matching coefficient. The larger the matching coefficient, the higher the consistency between the comprehensive respiratory rhythm feature and the trend index.

[0069] In practical applications, due to individual differences and the diversity of physiological states, there may be significant differences in the association strength between the respiratory rhythm and the changing trend of the exercise physiological state. To solve this problem, the system can adopt an adaptive weight optimization algorithm to dynamically adjust the weight coefficients of the respiratory rate and respiratory depth when synthesizing the comprehensive respiratory rhythm feature according to the physiological characteristics and exercise states of different users, improving the individual adaptability of the matching coefficient calculation. At the same time, the system can also introduce a multi-index decision fusion mechanism, comprehensively consider multiple correlation indexes such as the matching coefficient and correlation coefficient, and obtain a more reliable and robust association measure between the respiratory rhythm and the changing trend of the exercise physiological state through weighted combination or voting strategies.

[0070] S205. When the matching coefficient is greater than the preset synchronization threshold, increase the upper limit value of the preset reference interval by a preset first percentage and decrease the lower limit value by a preset second percentage; In this step, the system dynamically adjusts the upper and lower limit values of the preset reference interval according to the comparison result between the matching coefficient and the preset synchronization threshold. When the matching coefficient is greater than the preset synchronization threshold, it indicates that the user's respiratory rhythm and the changing trend of the exercise physiological state have a high degree of consistency, and the respiratory rhythm can better indicate the exercise health state. In this case, the system increases the upper limit value of the preset reference interval by a preset first percentage and at the same time decreases the lower limit value by a preset second percentage, expanding the range of the reference interval.

[0071] To implement this step, the system needs to preset appropriate synchronization thresholds and adjustment percentages in advance. The synchronization threshold can be obtained through statistical analysis of a large amount of population data or set by expert experience. The adjustment percentage needs to ensure the evaluation sensitivity while avoiding a decrease in stability caused by overly large adjustments. When adjusting the reference interval, the system first determines whether the matching coefficient is greater than the synchronization threshold. If the condition is met, the system multiplies the upper limit value of the reference interval by (1 + the first percentage) and multiplies the lower limit value by (1 - the second percentage) to obtain the adjusted reference interval.

[0072] S206. When the matching coefficient is less than the preset synchronization threshold, reduce the upper limit value of the preset reference interval by the preset first percentage and increase the lower limit value by the preset second percentage.

[0073] In this step, the system dynamically adjusts the upper and lower limit values of the preset reference interval according to the comparison result between the matching coefficient and the preset synchronization threshold. When the matching coefficient is less than the preset synchronization threshold, it indicates that the consistency between the user's breathing rhythm and the changing trend of the exercise physiological state is low, and the breathing rhythm has limited indication of the exercise health state. In this case, the system reduces the upper limit value of the preset reference interval by the preset first percentage, and at the same time increases the lower limit value by the preset second percentage to narrow the range of the reference interval. The specific implementation method is similar to the previous step and will not be elaborated here.

[0074] In the above embodiments, by analyzing the distribution of the peaks and valleys of the exercise physiological parameter correlation curve to determine the exercise cycle characteristics, segmenting the trend index based on this, and performing breathing rhythm analysis, the exercise intensity and breathing state can be correlated and analyzed. By calculating the matching coefficient of the breathing frequency, breathing depth and the trend index, the upper and lower limit values of the preset reference interval are dynamically adjusted, enabling the system to adaptively adjust the evaluation criteria according to the user's breathing state. When the breathing rhythm is highly matched with the exercise intensity, the reference interval is expanded, otherwise it is narrowed, improving the personalization and accuracy of exercise health detection and making the detection results more in line with the user's actual physiological state.

[0075] Further, in another embodiment, the system obtains the peak points and valley points of multiple adjacent exercise cycles; Calculate the peak slope between adjacent peak points and the valley slope between adjacent valley points; Determine the fatigue compensation coefficient according to the ratio of the peak slope to the valley slope, specifically including: When the slope of the peak point is positive and the slope of the valley point is negative, set the fatigue compensation coefficient to the first preset value; When the slope of the peak point is negative and the slope of the valley point is negative, set the fatigue compensation coefficient to the second preset value; When the slope of the peak point is negative and the slope of the valley point is positive, set the fatigue compensation coefficient to the third preset value, where the first preset value is greater than the second preset value, and the second preset value is greater than the third preset value; Multiply the matching coefficient by the fatigue compensation coefficient to obtain the corrected matching coefficient.

[0076] The system first obtains the peak points and valley points within multiple adjacent motion cycles, and then calculates the peak slope between adjacent peak points and the valley slope between adjacent valley points. The peak slope and the valley slope reflect the speed and trend of the change in the physiological state during the movement.

[0077] Next, the system determines the fatigue compensation coefficient based on the ratio of the peak slope to the valley slope. The fatigue compensation coefficient is used to quantify the impact of the user's fatigue level on the assessment of the exercise health status. Specifically, when the peak slope is positive and the valley slope is negative, it indicates that the user's physiological state shows an upward trend and the fatigue level is relatively low. At this time, set the fatigue compensation coefficient to the larger first preset value. When the peak slope is negative and the valley slope is negative, it indicates that the user's physiological state shows a downward trend and the fatigue level has increased. At this time, set the fatigue compensation coefficient to the smaller second preset value. When the peak slope is negative and the valley slope is positive, it indicates that the user's physiological state shows an obvious downward trend and the fatigue level is relatively high. At this time, set the fatigue compensation coefficient to the smallest third preset value.

[0078] Finally, the system multiplies the matching coefficient by the fatigue compensation coefficient to obtain the corrected matching coefficient. The corrected matching coefficient comprehensively considers the synchronization degree between the respiratory rhythm and the change trend of the exercise physiological state, as well as the user's fatigue level, and can more comprehensively and accurately evaluate the user's exercise health status.

[0079] To implement this embodiment, the system needs to design a reasonable calculation method and preset values for the fatigue compensation coefficient. The calculation method of the fatigue compensation coefficient can be based on the statistical analysis of a large amount of population data, or it can be combined with the empirical knowledge of sports physiology and medical experts. The setting of the preset values needs to take into account the compensation effects under different fatigue levels, reflecting the differences in fatigue levels while avoiding excessive or insufficient compensation amplitudes. When calculating the corrected matching coefficient, the system can use weighted multiplication, assigning different weights to the matching coefficient and the fatigue compensation coefficient to balance their contribution degrees.

[0080] In the above embodiments, by analyzing the change characteristics of the peak points and valley points in adjacent motion cycles and calculating the ratio of the peak slope to the valley slope to determine the fatigue compensation coefficient, the fatigue state during exercise can be effectively identified. The peak slope reflects the change trend of the exercise intensity, and the valley slope reflects the change trend of the recovery ability. The ratio of the two can characterize the dynamic balance relationship between exercise and recovery. Multiplying the matching coefficient by the fatigue compensation coefficient to obtain the corrected matching coefficient introduces the influence of the fatigue factor on the basis of the original matching evaluation of the breathing rhythm and the exercise trend, making the evaluation of the exercise health state more comprehensive and accurate. The corrected matching coefficient can better reflect the exercise adaptation ability of the user under different fatigue degrees, which helps to timely detect potential exercise risks.

[0081] The following describes the system in the embodiments of the present invention application from the perspective of hardware processing. Please refer to Figure 3 , which is a schematic structural diagram of an entity device of an AI intelligent exercise health detection system based on images provided in the embodiments of the present application.

[0082] It should be noted that Figure 3 the structure of the system shown is only an example and should not bring any limitations to the functions and usage scopes of the embodiments of the present invention.

[0083] As Figure 3 shown, the system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage section 308 into the random access memory (RAM) 303, such as executing the methods in the above embodiments. In the RAM 303, various programs and data required for system operation are also stored. The CPU 301, ROM 302, and RAM 303 are connected to each other through a bus 304. The input / output (I / O) interface 305 is also connected to the bus 304.

[0084] The following components are connected to the I / O interface 305: an input section 306 including a camera, an infrared sensor, etc.; an output section 307 including a liquid crystal display (LCD), a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc. is installed on the drive 310 as needed so that a computer program read therefrom is installed into the storage section 308 as needed.

[0085] Specifically, according to an embodiment of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, an embodiment of the present invention includes a computer program product that includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for performing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 309, and / or installed from the removable medium 311. When the computer program is executed by a central processing unit (CPU) 301, various functions defined in the present invention are executed.

[0086] It should be noted that the computer-readable medium shown in the embodiments of the present invention can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of a computer-readable storage medium can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable computer program. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above.

[0087] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present invention. Among them, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code, and the above module, program segment, or part of code contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, and the combination of blocks in the block diagram or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0088] As another aspect, the present invention also provides a computer-readable storage medium, which may be included in the system described in the above embodiments; or may exist alone without being assembled into the system. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of a system, the system implements the method provided in the above embodiments.

[0089] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

[0090] As used in the above embodiments, depending on the context, the term "when..." can be interpreted to mean "if...", or "after...", or "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if detecting (the stated condition or event)" can be interpreted to mean "if determining...", or "in response to determining...", or "when detecting (the stated condition or event)", or "in response to detecting (the stated condition or event)".

[0091] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center in a wired (such as coaxial cable, optical fiber, digital subscriber line) or wireless (such as infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that can be accessed by a computer, or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid-state drive), etc.

[0092] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above-described embodiments can be completed by hardware instructed by a computer program. This program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-described method embodiments. The aforementioned storage media include: various media such as ROM, random access memory (RAM), magnetic disks, or optical discs that can store program codes.

Claims

1. An image-based AI intelligent sports health detection method, characterized in that: include: Extracting the coordinate data of key points of the human body and the facial feature areas from the continuous video images captured by the camera during the user's movement; Calculating the displacement change rate of the key points based on the coordinate data of the key points of the human body, and obtaining the motion speed parameter according to the displacement change rate; Perform skin color analysis on the facial feature area to extract color feature parameters, wherein the color feature parameters include a red component ratio of the user's facial area; Performing temporal correlation between the motion speed parameter and the color characteristic parameter at the corresponding time point to obtain a motion physiological parameter correlation curve; Segmenting the sports physiological parameter association curve according to a preset time window, and calculating the parameter fluctuation range of each segment of the sports physiological parameter association curve; Determine a change trend between parameter fluctuation ranges of adjacent preset time windows, and generate a trend index based on the change trend; The trend index is compared with a preset reference interval to obtain a comparison result, and a sports health detection report is generated based on the comparison result.

2. The method according to claim 1, characterized in that The method of calculating the displacement change rate of the key points based on the coordinate data of the key points of the human body, and obtaining the motion speed parameter according to the displacement change rate, specifically includes: Establishing a relative motion coordinate system with the center of the user's trunk as the origin, and mapping the coordinate data of the key points of the human body to the relative motion coordinate system; Calculating the three-dimensional displacement vectors of the corresponding key points in the continuous video images of adjacent frames, and obtaining the displacement change rate of each key point according to the position change of each key point in the relative motion coordinate system; The displacement change rate of each key point is converted into a motion speed parameter according to the acquisition frame rate of the camera device.

3. The method according to claim 1, characterized in that: The determining of the variation trend between the parameter fluctuation ranges of adjacent preset time windows and generating a trend index based on the variation trend specifically includes: Calculating the difference in parameter fluctuation ranges of the sports physiological parameter association curves of adjacent preset time windows to obtain a fluctuation range difference value; Dividing the fluctuation range difference value by the time interval between adjacent preset time windows to obtain the change rate of the parameter fluctuation range; Determine the change trend of the sports physiological parameter association curve according to the positive or negative of the change rate, wherein a positive change rate indicates an upward trend, and a negative change rate indicates a downward trend; Performing weighted calculation on the rate of change of the parameter fluctuation range based on the change trend to obtain a weighted rate of change; The weighted change rate is mapped into a preset interval to generate a trend index.

4. The method according to claim 1, characterized in that After comparing the trend index with the preset reference interval to obtain a comparison result, and generating a sports health detection report based on the comparison result, the method further includes: Determine the peak and trough distribution of the sports physiological parameter correlation curve; Determining the motion cycle characteristics of the user according to the peak and trough distribution; The trend index is segmented based on the motion cycle characteristics, and respiratory rhythm analysis is performed on each segment of the trend index to obtain respiratory frequency and respiratory depth; Fitting the respiratory frequency and the respiratory depth with the trend index to obtain a matching coefficient; When the matching coefficient is greater than a preset synchronization threshold, increasing the upper limit value of the preset reference interval by a preset first percentage and decreasing the lower limit value by a preset second percentage; When the matching coefficient is less than the preset synchronization threshold, the upper limit value of the preset reference interval is reduced by the preset first percentage and the lower limit value is increased by the preset second percentage.

5. The method according to claim 4, characterized in that The step of fitting the respiratory frequency and the respiratory depth with the trend index to obtain a matching coefficient specifically includes: Recording the time series curve of the respiratory frequency as a first curve, recording the time series curve of the respiratory depth as a second curve, and recording the time series curve of the trend index as a third curve; respectively calculating the Pearson correlation coefficient between the first curve and the third curve and the Pearson correlation coefficient between the second curve and the third curve; Normalizing the first curve and the second curve to obtain a normalized curve; The normalized curves are synthesized into a comprehensive curve by weighted summation; The Euclidean distance between the comprehensive curve and the third curve is calculated, and the reciprocal of the Euclidean distance is used as a matching coefficient.

6. The method according to claim 4, characterized in that After lowering the upper limit of the preset reference interval by the preset first percentage and increasing the lower limit by the preset second percentage, the method further includes: Obtain peak points and valley points of multiple adjacent motion cycles; Calculating the peak slope between adjacent peak points and the valley slope between adjacent valley points; Determining a fatigue compensation coefficient according to a ratio of the peak slope to the valley slope; The matching coefficient is multiplied by the fatigue compensation coefficient to obtain a corrected matching coefficient.

7. The method according to claim 6, characterized in that Determining the fatigue compensation coefficient according to the ratio of the peak slope to the valley slope specifically includes: When the peak point slope is positive and the valley point slope is negative, setting the fatigue compensation coefficient to a first preset value; When the peak point slope is negative and the valley point slope is negative, setting the fatigue compensation coefficient to a second preset value; When the peak point slope is negative and the valley point slope is positive, the fatigue compensation coefficient is set to a third preset value, the first preset value is greater than the second preset value, and the second preset value is greater than the third preset value.

8. An image-based AI intelligent sports health detection system, characterized in that: The system comprises: One or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the system to execute the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a system, the system is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on a system, the system is caused to execute the method according to any one of claims 1 to 7.

Images