Unmanned aerial vehicle-based human body motion posture correction method, system and equipment and medium

Abstract

The invention discloses a human motion posture correction method, system and device based on an unmanned aerial vehicle and a medium, and the method comprises the steps: obtaining a load training scheme of a user through clustering analysis, and determining a posture guidance scheme; judging whether the user is in a normal movement mode or not when the user moves; if not, reminding the user to pause the exercise; if yes, shooting a continuous frame video stream in the motion of the user; dividing the video stream according to the minimum action unit to obtain an image, extracting user contour features and calculating ambiguity, and performing deblurring to obtain a target contour so as to extract key points of a user skeleton; and according to the ideal coordinates and the actual coordinates of the skeleton key points, calculating a joint angle deviation in the motion attitude of the image frame so as to calculate a target angle deviation of the continuous motion attitude of the user in the minimum motion unit, and updating a load training scheme and an optimized attitude guidance scheme of the user. The training plan of the user can be dynamically adjusted, the load bearing scheme of the user is scientifically guided, posture correction is carried out, and the training effect is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of motion monitoring, and in particular to a method, system, equipment and medium for correcting human motion posture based on an unmanned aerial vehicle. Background Art

[0002] Sports weight training is a form of physical exercise that is performed by adding external loads. During weight training, athletes or fitness enthusiasts carry additional weights such as sandbags, vests, etc. to perform various forms of exercises, thereby accelerating the training effect and enhancing muscle strength.

[0003] Current weight training programs are often based on static data and lack the ability to adjust in real time, resulting in the possibility that the training plan may not meet the actual needs of the individual. During the training process of athletes, unreasonable weight-bearing programs will further affect the movement posture and increase the risk of sports injuries. In order to monitor the movement process, traditional posture correction systems usually only rely on fixed equipment, and there are problems with low accuracy and poor flexibility in posture recognition, which leads to unsatisfactory correction effects. Summary of the invention

[0004] In order to solve at least one of the technical problems mentioned above, the present invention provides a method, system, device and medium for correcting human motion posture based on an unmanned aerial vehicle.

[0005] In a first aspect, the present invention provides a method for correcting human motion posture based on an unmanned aerial vehicle, the method comprising:

[0006] Collecting first physiological data of the user before exercise, clustering the first physiological data with data from a sports guidance program case library, generating a weight training program for the user, and determining a posture guidance program based on the weight training program;

[0007] The sensor collects the second physiological data of the user during exercise, and determines whether the user is in a normal exercise mode according to the second physiological data; if not, remind the user to stop exercising; if yes, use a drone to shoot continuous frame video stream data of the user during exercise;

[0008] The video stream is divided into multiple groups of continuous frame images according to the minimum action unit. Based on each group of continuous frame images, the corresponding user contour features are extracted and the fuzziness of the features is calculated. The deblurring operation is matched according to the fuzziness to generate the target contour features, and the user skeleton key points are extracted from the target contour features of the continuous frames.

[0009] Determine the actual coordinates of the user's skeletal key points in the continuous frame images and the ideal coordinates of the skeletal key points in the posture guidance scheme, and calculate the joint angle deviations of different skeletal key points of the user in the image frame motion posture according to the actual coordinates and the ideal coordinates;

[0010] The target angle deviation of the user's continuous motion posture in the minimum action unit is calculated based on multiple joint angle deviations, the user's weight training plan is updated according to the target angle deviation, and the updated weight training plan is used to optimize the posture guidance plan.

[0011] Preferably, updating the user's weight training program according to the target angle deviation includes:

[0012] Suppose a joint consists of three key points A, B, and C. The coordinates of A, B, and C are (x A ,y A ), (x B ,y B ), (x C ,y C ), determine the vector and

[0013] calculate

[0014] calculate

[0015]

[0016] Determine θ according to the calculation formula of cosθ ij，a ,θ ij，g , then:

[0017] Δθ tj =|θ ij，a -θ ij，g |;

[0018] In the formula, Δθ tj represents the angle deviation of the i-th joint in the j-th frame image, θ ij，a ,θ ij，g They represent the actual angle and ideal angle of the i-th joint in the j-th frame image respectively;

[0019] Assuming there are m joints and n frames of images, determine the target angle deviation of the user's continuous motion posture in the minimum action unit

[0020]

[0021] Get the average speed of the i-th joint in n frames of images and the average acceleration

[0022]

[0023] In the formula, v ij 、aij Respectively represent the velocity and acceleration of the i-th joint in the j-th frame image;

[0024] To update a user's weight training program:

[0025]

[0026] Where W and W0 are the training weights before and after the user update, respectively; k is an adjustment factor with a value range of (0.01, 0.1); G and G0 are the user weight and the preset standard weight, respectively; and e is a natural constant.

[0027] Preferably, the collecting of the first physiological data of the user before exercise, clustering the first physiological data with the data of the exercise guidance program case library, and generating the weight training program for the user includes:

[0028] Acquiring first physiological data, including the user's age, height, weight, and heart rate, blood pressure, EMG signal strength, and blood oxygen concentration before exercise;

[0029] Performing data cleaning and normalization operations on the first physiological data, and using the PCA algorithm to perform feature dimension reduction on the normalized data;

[0030] Analyze the correlation between the features after dimensionality reduction and the load parameters, remove the features whose correlation is less than the preset threshold, and use the remaining features as the target features;

[0031] The K-Means clustering algorithm is used to cluster the target features and the data features of the exercise guidance program case library to generate the user's weight training plan.

[0032] In a second aspect, the present invention further provides a human motion posture correction system based on an unmanned aerial vehicle, the system comprising:

[0033] An exercise program clustering unit, used for collecting first physiological data of the user before exercise, clustering the first physiological data with data of the exercise guidance program case library, generating a weight training program for the user, and determining a posture guidance program according to the weight training program;

[0034] A motion video acquisition unit, used to collect the second physiological data of the user during motion based on the sensor, and determine whether the user is in a normal motion mode according to the second physiological data; if not, remind the user to stop exercising; if yes, use a drone to shoot continuous frame video stream data during the user's motion;

[0035] A skeleton key point extraction unit is used to divide the video stream into multiple groups of continuous frame images according to the minimum action unit, extract the corresponding user contour features based on each group of continuous frame images and calculate the fuzziness of the features, match the deblurring operation according to the fuzziness to generate the target contour features, and extract the user skeleton key points from the target contour features of the continuous frames;

[0036] An angle deviation calculation unit, used to determine the actual coordinates of the user's skeleton key points in the continuous frame images and the ideal coordinates of the skeleton key points in the posture guidance scheme, and calculate the joint angle deviations of different skeleton key points of the user in the image frame motion posture according to the actual coordinates and the ideal coordinates;

[0037] The motion plan optimization unit is used to calculate the target angle deviation of the user's continuous motion posture in the minimum action unit based on multiple joint angle deviations, update the user's weight training plan based on the target angle deviation, and optimize the posture guidance plan using the updated weight training plan.

[0038] Preferably, the motion scheme optimization unit is further used for:

[0039] Suppose a joint consists of three key points A, B, and C. The coordinates of A, B, and C are (x A ,y A ), (x B ,y B ), (x C ,y C ), determine the vector and

[0040] calculate

[0041] calculate

[0042]

[0043] Determine θ according to the calculation formula of cosθ ij，a ,θ ij，g , then:

[0044] Δθ ij =|θ ij，a -θ ij，g |;

[0045] In the formula, Δθ ij represents the angle deviation of the i-th joint in the j-th frame image, θ ij，a ,θ ij，g They represent the actual angle and ideal angle of the i-th joint in the j-th frame image respectively;

[0046] Assuming there are m joints and n frames of images, determine the target angle deviation of the user's continuous motion posture in the minimum action unit

[0047]

[0048] Get the average speed of the i-th joint in n frames of images and the average acceleration

[0049]

[0050] In the formula, v ij 、a ij Respectively represent the velocity and acceleration of the i-th joint in the j-th frame image;

[0051] To update a user's weight training program:

[0052]

[0053] Where W and W0 are the training weights before and after the user update, respectively; k is an adjustment factor with a value range of (0.01, 0.1); G and G0 are the user weight and the preset standard weight, respectively; and e is a natural constant.

[0054] Preferably, the motion scheme clustering unit is further used for:

[0055] Acquiring first physiological data, including the user's age, height, weight, and heart rate, blood pressure, EMG signal strength, and blood oxygen concentration before exercise;

[0056] Performing data cleaning and normalization operations on the first physiological data, and using the PCA algorithm to perform feature dimension reduction on the normalized data;

[0057] Analyze the correlation between the features after dimensionality reduction and the load parameters, remove the features whose correlation is less than the preset threshold, and use the remaining features as the target features;

[0058] The K-Means clustering algorithm is used to cluster the target features and the data features of the exercise guidance program case library to generate the user's weight training plan.

[0059] In a third aspect, the present invention further provides an electronic device comprising a processor and a memory, wherein the memory is used to store computer program code, and the computer program code comprises computer instructions. When the processor executes the computer instructions, the electronic device executes the method as described in the first aspect above and any possible implementation thereof.

[0060] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, wherein the computer program includes program instructions, and when the program instructions are executed by a processor of an electronic device, the processor executes the method as described in the first aspect above and any possible implementation thereof.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] 1) By combining cluster analysis with the user's first physiological data, a personalized weight training plan can be provided for the user to ensure that the training intensity is in line with the individual's health status and physical fitness level, making the generated posture guidance plan more scientific and reasonable, helping users to master the correct exercise posture and reduce the risk of sports injuries caused by improper posture.

[0063] 2) By using sensors to monitor the user's physiological state in real time, such as heart rate and blood oxygen saturation, the user's exercise intensity can be understood in a timely manner. When the user is not in normal exercise mode, the system will remind them to stop exercising to prevent overwork or injury. Drone shooting is only started when the user is in normal exercise mode, ensuring effective acquisition of exercise training data and enhancing the safety monitoring efficiency during training.

[0064] 3) By dividing the video into the smallest action units, the details of each action can be analyzed more carefully. The deblurring operation improves the image quality, ensures the clarity of the contour features, improves the efficiency and accuracy of data processing, and thus more accurately extracts the key points of the skeleton, greatly increasing the accuracy of motion posture recognition.

[0065] 4) By comparing the actual coordinates with the ideal coordinates, it is possible to accurately assess whether the user's exercise posture meets the expected standard. By quantifying the joint angle deviation and further calculating the target angle deviation, the user's training plan can be dynamically adjusted to make it more in line with the individual's progress. This can continuously optimize the posture guidance program, ensure that users can always get the most effective guidance, help users better master the correct exercise skills, and greatly improve the training effect.

[0066] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the background technology, the drawings required for use in the embodiments of the present invention or the background technology will be described below.

[0068] The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and are used to illustrate the technical solutions of the present disclosure together with the specification.

[0069] Figure 1 A schematic diagram of a flow chart of a method for correcting human motion posture based on a drone provided in an embodiment of the present invention;

[0070] Figure 2 yes Figure 1 Schematic diagram of the process of the sub-steps of step S10;

[0071] Figure 3 A schematic diagram of the structure of a human motion posture correction system based on an unmanned aerial vehicle provided in an embodiment of the present invention;

[0072] Figure 4 A schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0073] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0074] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0075] See also Figure 1 , Figure 1 The present invention provides a flowchart of a method for correcting human motion posture based on a drone. Figure 1 As shown, the method includes:

[0076] S10, collecting the first physiological data of the user before exercise, clustering the first physiological data with the data of the exercise guidance program case library, generating a weight training program for the user, and determining a posture guidance program according to the weight training program.

[0077] Specifically, see Figure 2 , Figure 2 is a schematic diagram of the sub-steps of step S10. Figure 2 As shown, the collecting of the first physiological data of the user before exercise, clustering the first physiological data with the data of the exercise guidance program case library, and generating the weight training program for the user include:

[0078] S101, obtaining first physiological data, including the user's age, height, weight, and heart rate, blood pressure, EMG signal strength, and blood oxygen concentration before exercise;

[0079] S102, performing data cleaning and normalization operations on the first physiological data, and using a PCA algorithm to perform feature dimension reduction on the normalized data;

[0080] S103, analyzing the correlation between the features after dimensionality reduction and the load parameters, eliminating the features whose correlation is less than a preset threshold, and taking the remaining features as target features;

[0081] S104. Use the K-Means clustering algorithm to cluster the target features and the data features of the exercise guidance program case library to generate a weight training program for the user.

[0082] In this embodiment, the user's age, height, and weight are usually registered before training to build an identity file, and data such as face or fingerprint information is stored in the database. Before the user receives training, the current user's age, height, and weight can be quickly read through face recognition technology or fingerprint recognition. The physiological indicators such as heart rate, blood pressure, EMG signal strength, and blood oxygen concentration before exercise can be measured by devices such as smart bracelets, watches, sphygmomanometers, and electromyographs (EMG). Before measurement, it is necessary to ensure that all devices are connected and synchronized to the data processing platform, and start recording the user's physiological data for a period of time before preparing for exercise. By collecting a variety of physiological data, it is helpful to comprehensively evaluate the user's health status and physical fitness level, and provide necessary physiological parameter support for the subsequent generation of personalized training programs.

[0083] After obtaining the first physiological data, invalid or abnormal data points can be removed through data preprocessing, missing values ​​can be filled, and noise interference can be eliminated. Then the data can be converted to the same scale through standardization or normalization methods. The principal component analysis (PCA) algorithm is then used to reduce the dimension of the normalized data, retain the main features, and reduce the data dimension. The quality of the data is improved by cleaning and normalization operations, and the impact of noise on subsequent analysis is reduced. The amount of calculation is reduced and the processing speed is improved by dimensionality reduction. After obtaining the reduced-dimensional features, the correlation coefficient between them and the weight-bearing parameters (such as weight, number of repetitions, etc.) is calculated. When performing feature screening, this embodiment sets a threshold to eliminate features with correlations lower than this threshold, and only retains those features that are highly correlated with the weight-bearing parameters as target features. In this way, insignificant features can be removed, the model can be simplified to avoid overfitting, and it is ensured that the features ultimately used for clustering are directly related to weight-bearing training, thereby improving the effectiveness of the scheme.

[0084] Finally, the K-Means algorithm is used to cluster the target features and find cases with similar features to the user. It is understandable that the sports guidance case library can provide different scientific guidance plans for users of different heights, ages, weights, and physiological indicators to ensure better training effects and higher training safety. After cluster analysis, combined with the results of the clustering cluster, the case closest to the user's features can be selected, so that the user's weight training plan can be quickly generated with reference to the case.

[0085] Once you have a scientific weight training program, you can identify typical movement patterns related to weight training. For example, if it is leg strength training, it may involve squats, deadlifts, etc.; if it is upper limb training, it may include bench presses, pull-ups, etc. For each training movement, define the ideal posture standard, such as:

[0086] Starting position: The position the user should assume when beginning the movement.

[0087] Motion path: The movement trajectory of each joint when the user performs an action.

[0088] End position: The position the user should be in after the movement is completed.

[0089] Therefore, this embodiment uses cluster analysis to find the training plan that best suits the user's characteristics, implements personalized recommendations, ensures that the generated training plan is closer to the user's actual situation, improves the scientificity and effectiveness of the training, helps users master the correct exercise posture, and reduces the risk of sports injuries caused by improper posture.

[0090] S20, collecting the second physiological data of the user during exercise based on the sensor, and determining whether the user is in a normal exercise mode according to the second physiological data; if not, reminding the user to stop exercising; if so, using a drone to shoot continuous frame video stream data of the user during exercise.

[0091] In this embodiment, sensors such as a heart rate monitor, an accelerometer, and a gyroscope can be integrated into the user's wearable device (such as a smart watch or a fitness band). These sensors collect the user's physiological data in real time, such as heart rate, movement frequency and intensity, etc. According to the second physiological data center frequency, blood pressure, EMG signal strength, and blood oxygen concentration value, it is judged whether the user is in a normal exercise mode. For example, the safety thresholds of these indicators can be set first. Once it is found that a certain indicator exceeds the safety threshold, it can be determined that it is in an abnormal exercise mode and the user needs to be reminded to suspend the exercise mode. If these indicators are within their respective safety thresholds, a comprehensive indicator score can also be set, and the comprehensive exercise score is determined in the form of weighted summation, and it is compared with the exercise safety threshold set in advance under the normal exercise mode. Within this range, it means that the user is in a normal exercise mode, otherwise it means that the user may feel physically uncomfortable and needs to stop exercising immediately. Preferably, the model can also be trained by a machine learning algorithm to identify the normal exercise mode. The second physiological data collected in real time is input into the trained model for analysis to determine whether the current exercise conforms to the normal mode. If the algorithm detects that the user's physiological data deviates from the normal range, it is considered that the user may be in an abnormal exercise state. Once an abnormal situation is detected, the system will send an alarm or prompt to the user through the device, suggesting that he or she stop exercising and take a rest.

[0092] Furthermore, if the user's movement conforms to the normal mode, the paired drone can be started for tracking shooting. The drone transmits the continuous frame video stream data of the user's movement process to the user's mobile device or cloud storage in real time. When the drone shoots the user's movement video, it can collect video data at different heights and angles, so as to more effectively identify the human body's movement posture. Therefore, this embodiment can timely detect potential physical discomfort or overwork by monitoring the user's physiological data in real time, thereby preventing the occurrence of sports injuries. By providing users with personalized exercise suggestions, they can help them adjust their exercise plans to achieve the best health benefits. In addition, the drone shooting is only started when the user is in a normal exercise state to ensure the effective acquisition of sports training data and enhance the safety monitoring efficiency during the training process.

[0093] S30, dividing the video stream into multiple groups of continuous frame images according to the minimum action unit, extracting the corresponding user contour features based on each group of continuous frame images and calculating the blurriness of the features, matching the deblurring operation according to the blurriness to generate the target contour features, and extracting the user skeleton key points from the target contour features of the continuous frames.

[0094] "Minimum action unit" refers to the most basic action component when completing a task or activity. In sports training engineering, for example, when lifting weights, a complete weight lifting action (from lifting the barbell from the ground to putting it down) can also be regarded as consisting of multiple minimum action units, such as squatting, lifting, standing, putting down, etc. When running, the smallest action unit can be a change from landing to pushing off the ground, for example, from left foot landing - right foot pushing off the ground - front drive running - right foot landing, which can be regarded as a minimum action unit. Furthermore, the video stream can be divided into multiple groups of continuous frame images according to the minimum action unit, and each group of continuous frame images fully records the continuous movement posture of a minimum action unit.

[0095] Furthermore, based on each set of continuous frame images, the corresponding user contour features are extracted and the fuzziness of the features is calculated. The deblurring operation is matched according to the fuzziness to generate the target contour features, specifically:

[0096] Image processing: Remove image noise to improve the accuracy of subsequent processing, and then convert the color image to grayscale to simplify the complexity of subsequent processing. Perform histogram equalization to improve image contrast and make edges easier to detect.

[0097] User segmentation: Use background subtraction methods such as Gaussian mixture models to distinguish foreground and background, compare the differences between adjacent frames to detect moving objects, and use edge detection algorithms such as Canny edge detection to identify user outlines.

[0098] Feature extraction: Extract closed contours from edge detection results and identify key points on the contours, such as corners, endpoints, etc. Then use feature descriptors to calculate descriptors around feature points for subsequent matching or classification.

[0099] Contour matching and tracking: Use template matching or other similarity measurement methods to determine the contour of the same user in consecutive frames. Use optical flow or other tracking algorithms to continuously track the position changes of the user in multiple frames. Finally, reorganize the extracted contour features into meaningful information, such as human posture recognition. Convert contour features into numerical feature vectors for subsequent applications.

[0100] After the contour features are obtained, in order to improve the clarity of the contour features and provide more accurate data for subsequent gesture recognition, this embodiment will perform fuzziness evaluation and optimization, specifically:

[0101] 1) Calculate the motion blur kernel:

[0102] Initialize the blur radius R, and use the blur radius R to construct a motion vector diagram, where the motion vector diagram is a straight line with a length of R, an angle of 0, and a width of 1 pixel;

[0103] Based on the blur radius R, establish the affine transformation matrix K:

[0104]

[0105] In the formula, θ is the moving angle of the UAV;

[0106] The motion vector map is mapped using the affine transformation matrix K to obtain a motion blur kernel.

[0107] 2) Calculate the ambiguity:

[0108] The motion blur kernel is used to determine the pixel value of each point in the contour feature, and the blur degree B of the video image is calculated according to the pixel value, specifically:

[0109]

[0110] Where L is the width of the image, H is the height of the image, and P lh is the pixel value of the image at pixel (l, h).

[0111] 3) Fuzziness tuning:

[0112] Determine whether the blur reaches the preset threshold. If the blur does not reach the preset value, multiply the blur radius by the compensation coefficient f, and then bring it back to the affine transformation matrix to recalculate the motion blur kernel and blur until the blur reaches the preset value; where the compensation coefficient f ranges from 1≤f≤1.3.

[0113] After the blur is adjusted, the target contour features can be obtained, and the user's skeleton key points can be extracted from the target contour features of continuous frames. That is, the main structural lines of the human body. Key points are detected based on the skeleton, such as the head, shoulders, elbows, wrists, hips, knees and ankles, etc., and 18 key points are usually collected. Preferably, the human posture estimation framework OpenPose can be used to directly extract the coordinates of the key points of the human body from the image.

[0114] S40, determining the actual coordinates of the user's skeletal key points in the continuous frame images and the ideal coordinates of the skeletal key points in the posture guidance scheme, and calculating the joint angle deviations of different skeletal key points of the user in the image frame motion posture according to the actual coordinates and the ideal coordinates.

[0115] In this embodiment, a set of key point coordinates in a standard posture is defined according to the type of sport. These coordinates can be based on the standard posture of professional athletes, or the best practice posture for a specific sport. The key point coordinates of the standard posture are standardized to adapt to different individual sizes, such as by scaling. The actual coordinates are then normalized to adapt to the scale of the ideal coordinates. The actual coordinates are mapped to the same reference system as the ideal coordinates.

[0116] Further, the joints to be analyzed (such as shoulder joints, elbow joints, hip joints, etc.) are selected. The actual angles of each joint are calculated based on the actual coordinates. Trigonometric functions are usually used, such as the angle formula of vectors. Finally, the actual angle is compared with the ideal angle to calculate the angle deviation.

[0117] S50. Calculate the target angle deviation of the user's continuous motion posture in the minimum action unit according to the multiple joint angle deviations, update the user's weight training program according to the target angle deviation, and optimize the posture guidance program using the updated weight training program.

[0118] Specifically, updating the user's weight training program according to the target angle deviation includes:

[0119] Suppose a joint consists of three key points A, B, and C. The coordinates of A, B, and C are (x A ,y A ), (x B ,y B ), (x C ,y C ), determine the vector and

[0120] calculate

[0121] calculate

[0122]

[0123] Determine θ according to the calculation formula of cosθ ij，a ,θ ij，g , then:

[0124] Δθ ij =|θ ij，a -θ ij，g |;

[0125] In the formula, Δθ ij represents the angle deviation of the i-th joint in the j-th frame image, θ ij，a ,θ ij，g They represent the actual angle and ideal angle of the i-th joint in the j-th frame image respectively;

[0126] Assuming there are m joints and n frames of images, determine the target angle deviation of the user's continuous motion posture in the minimum action unit

[0127]

[0128] Get the average speed of the i-th joint in n frames of images and the average acceleration

[0129]

[0130] In the formula, v ij 、a ij Respectively represent the velocity and acceleration of the i-th joint in the j-th frame image;

[0131] To update a user's weight training program:

[0132]

[0133] Where W and W0 are the training weights before and after the user update, respectively; k is an adjustment factor with a value range of (0.01, 0.1); G and G0 are the user weight and the preset standard weight, respectively; and e is a natural constant.

[0134] In the above weight update process, an exponential function is used to adjust the weight, taking into account the angle deviation and the ratio of the user's weight to the standard weight, as well as the influence of speed and acceleration. This ensures that the training weight can reflect the user's actual training effect while taking into account the user's physical load. By monitoring the joint angle deviation, the user's improper posture or movement can be discovered in time, and the training plan can be adjusted in time to reduce the risk of injury. By dynamically adjusting the training weight, the training can be made more scientific and reasonable, avoiding overtraining or undertraining.

[0135] Therefore, when the above-mentioned embodiment is updating the weight-bearing scheme and correcting the posture, the motion video is collected by the drone, and the video is divided into the smallest action unit, so that the details of each action can be analyzed more carefully. The image quality is improved by the deblurring operation, the clarity of the contour features is ensured, and the efficiency and accuracy of data processing are improved, so that the key points of the skeleton are extracted more accurately, and the accuracy of motion posture recognition is greatly increased. By comparing the actual coordinates of the joint nodes with the ideal coordinates, it is possible to accurately evaluate whether the user's motion posture meets the expected standards. By quantifying the joint angle deviation and further calculating the target angle deviation, the user's training plan can be dynamically adjusted to make it more in line with the individual's progress. In this way, the posture guidance plan can be continuously optimized to ensure that the user can always get the most effective guidance, help the user better master the correct sports skills, and greatly improve the training effect and safety.

[0136] See also Figure 3 In one embodiment, the present invention further provides a human motion posture correction system based on an unmanned aerial vehicle, the system comprising:

[0137] The exercise program clustering unit 100 is used to collect the first physiological data of the user before exercise, cluster the first physiological data with the data of the exercise guidance program case library, generate a weight training program for the user, and determine the posture guidance program according to the weight training program;

[0138] The motion video acquisition unit 200 is used to collect the second physiological data of the user during the motion based on the sensor, and determine whether the user is in a normal motion mode according to the second physiological data; if not, remind the user to stop the motion; if so, use a drone to shoot the continuous frame video stream data of the user during the motion;

[0139] The skeleton key point extraction unit 300 is used to divide the video stream into multiple groups of continuous frame images according to the minimum action unit, extract the corresponding user contour features based on each group of continuous frame images and calculate the fuzziness of the features, match the deblurring operation according to the fuzziness to generate the target contour features, and extract the user skeleton key points from the target contour features of the continuous frames;

[0140] Angle deviation calculation unit 400, used to determine the actual coordinates of the user's skeleton key points in the continuous frame images and the ideal coordinates of the skeleton key points in the posture guidance scheme, and calculate the joint angle deviations of different skeleton key points of the user in the image frame motion posture according to the actual coordinates and the ideal coordinates;

[0141] The motion plan optimization unit 500 is used to calculate the target angle deviation of the user's continuous motion posture in the minimum action unit based on multiple joint angle deviations, update the user's weight training plan based on the target angle deviation, and use the updated weight training plan to optimize the posture guidance plan.

[0142] In one embodiment, the motion scheme optimization unit 200 is further configured to:

[0143] Suppose a joint consists of three key points A, B, and C. The coordinates of A, B, and C are (x A ,y A ), (x B ,y B ), (x C ,y C ), determine the vector and

[0144] calculate

[0145] calculate

[0146]

[0147] Determine θ according to the calculation formula of cosθ ij，a ,θ ij，g , then:

[0148] Δθ ij =|θ ij，a -θ ij，g |;

[0149] In the formula, Δθ ij represents the angle deviation of the i-th joint in the j-th frame image, θ ij，a ,θ ij，g They represent the actual angle and ideal angle of the i-th joint in the j-th frame image respectively;

[0150] Assuming there are m joints and n frames of images, determine the target angle deviation of the user's continuous motion posture in the minimum action unit

[0151]

[0152] Get the average speed of the i-th joint in n frames of images and the average acceleration

[0153]

[0154] In the formula, v ij 、a ij Respectively represent the velocity and acceleration of the i-th joint in the j-th frame image;

[0155] To update a user's weight training program:

[0156]

[0157] Where W and W0 are the training weights before and after the user update, respectively; k is an adjustment factor with a value range of (0.01, 0.1); G and G0 are the user weight and the preset standard weight, respectively; and e is a natural constant.

[0158] In one embodiment, the motion scheme clustering unit 100 is further configured to:

[0159] Acquiring first physiological data, including the user's age, height, weight, and heart rate, blood pressure, EMG signal strength, and blood oxygen concentration before exercise;

[0160] Performing data cleaning and normalization operations on the first physiological data, and using the PCA algorithm to perform feature dimension reduction on the normalized data;

[0161] Analyze the correlation between the features after dimensionality reduction and the load parameters, remove the features whose correlation is less than the preset threshold, and use the remaining features as the target features;

[0162] The K-Means clustering algorithm is used to cluster the target features and the data features of the exercise guidance program case library to generate the user's weight training plan.

[0163] It can be understood that the functions or modules included in the system provided in this embodiment can be used to execute the method described in the above method embodiment. Its specific implementation can refer to the description of the above method embodiment. For the sake of brevity, it will not be repeated here.

[0164] The present invention also provides an electronic device, including a processor and a memory, wherein the memory is used to store computer program code, and the computer program code includes computer instructions. When the processor executes the computer instructions, the electronic device executes a method as described in any possible implementation manner.

[0165] The present invention also provides a computer-readable storage medium, in which a computer program is stored. The computer program includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes a method as described in any possible implementation manner.

[0166] See also Figure 4 , Figure 4 A schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention.

[0167] The electronic device 2 includes a processor 21, a memory 22, an input device 23, and an output device 24. The processor 21, the memory 22, the input device 23, and the output device 24 are coupled via a connector, and the connector includes various interfaces, transmission lines, or buses, etc., which are not limited in the embodiments of the present invention. It should be understood that in various embodiments of the present invention, coupling refers to mutual connection in a specific manner, including direct connection or indirect connection through other devices, for example, through various interfaces, transmission lines, buses, etc.

[0168] The processor 21 may be one or more graphics processing units (GPUs). When the processor 21 is a GPU, the GPU may be a single-core GPU or a multi-core GPU. Optionally, the processor 21 may be a processor group consisting of multiple GPUs, and the multiple processors are coupled to each other via one or more buses. Optionally, the processor may also be other types of processors, etc., which are not limited in the embodiments of the present invention.

[0169] The memory 22 can be used to store computer program instructions and various computer program codes including program codes for executing the scheme of the present invention. Optionally, the memory includes but is not limited to random access memory (RAM), read-only memory (ROM), erasable programmable read only memory (EPROM), or portable read only memory (CD-ROM), which is used for related instructions and data.

[0170] The input device 23 is used to input data and / or signals, and the output device 24 is used to output data and / or signals. The input device 23 and the output device 24 can be independent devices or an integrated device.

[0171] It is understandable that in the embodiment of the present invention, the memory 22 is not only used to store related instructions, and the embodiment of the present invention does not limit the specific data stored in the memory.

[0172] Understandably, Figure 4 Only a simplified design of an electronic device is shown. In practical applications, the electronic device may also include other necessary components, including but not limited to any number of input / output devices, processors, memories, etc., and all video analysis devices that can implement the embodiments of the present invention are within the protection scope of the present invention.

[0173] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0174] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. Those skilled in the art can also clearly understand that the descriptions of the various embodiments of the present invention have different focuses. For the convenience and brevity of description, the same or similar parts may not be repeated in different embodiments. Therefore, for parts not described or not described in detail in a certain embodiment, refer to the records of other embodiments.

[0175] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0176] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0177] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0178] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by 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 process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode. 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 may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital versatile disc (DVD)), or a semiconductor medium (eg, a solid state disk (SSD)).

[0179] A person skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by a computer program to instruct the relevant hardware, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-mentioned method embodiments. The aforementioned storage medium includes: a read-only memory (ROM) or a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

Claims

1. A method for correcting human motion posture based on an unmanned aerial vehicle, characterized in that: The method comprises: Collecting first physiological data of the user before exercise, clustering the first physiological data with data from a sports guidance program case library, generating a weight training program for the user, and determining a posture guidance program based on the weight training program; The sensor collects the second physiological data of the user during exercise, and determines whether the user is in a normal exercise mode according to the second physiological data; if not, remind the user to stop exercising; if yes, use a drone to shoot continuous frame video stream data of the user during exercise; The video stream is divided into multiple groups of continuous frame images according to the minimum action unit. Based on each group of continuous frame images, the corresponding user contour features are extracted and the fuzziness of the features is calculated. The deblurring operation is matched according to the fuzziness to generate the target contour features, and the user skeleton key points are extracted from the target contour features of the continuous frames. Determine the actual coordinates of the user's skeletal key points in the continuous frame images and the ideal coordinates of the skeletal key points in the posture guidance scheme, and calculate the joint angle deviations of different skeletal key points of the user in the image frame motion posture according to the actual coordinates and the ideal coordinates; The target angle deviation of the user's continuous motion posture in the minimum action unit is calculated based on multiple joint angle deviations, the user's weight training plan is updated according to the target angle deviation, and the updated weight training plan is used to optimize the posture guidance plan.

2. The method for correcting human motion posture based on an unmanned aerial vehicle according to claim 1, characterized in that: The updating of the user's weight training program according to the target angle deviation comprises: Suppose a joint consists of three key points A, B, and C. The coordinates of A, B, and C are (x A ,y A ), (x B ,y B ), (x C ,y C ), determine the vector and calculate calculate Determine θ according to the calculation formula of cosθ ij,a ,θ ij,g , then: Dth ij =|θ ij,a -θ ij,g |; In the formula, Δθ ij represents the angle deviation of the i-th joint in the j-th frame image, θ ij,a ,θ ij,g They represent the actual angle and ideal angle of the i-th joint in the j-th frame image respectively; Assuming there are m joints and n frames of images, determine the target angle deviation of the user's continuous motion posture in the minimum action unit Get the average speed of the i-th joint in n frames of images and the average acceleration In the formula, v ij 、a ij Respectively represent the velocity and acceleration of the i-th joint in the j-th frame image; To update a user's weight training program: Where W and W0 are the training weights before and after the user update, respectively; k is an adjustment factor with a value range of (0.01, 0.1); G and G0 are the user weight and the preset standard weight, respectively; and e is a natural constant.

3. The method for correcting human motion posture based on an unmanned aerial vehicle according to claim 1, characterized in that: The collecting of the first physiological data of the user before exercise, clustering the first physiological data with the data of the exercise guidance program case library, and generating the weight training program for the user includes: Acquiring first physiological data, including the user's age, height, weight, and heart rate, blood pressure, EMG signal strength, and blood oxygen concentration before exercise; Performing data cleaning and normalization operations on the first physiological data, and using the PCA algorithm to perform feature dimension reduction on the normalized data; Analyze the correlation between the features after dimensionality reduction and the load parameters, remove the features whose correlation is less than the preset threshold, and use the remaining features as the target features; The K-Means clustering algorithm is used to cluster the target features and the data features of the exercise guidance program case library to generate the user's weight training plan.

4. A human motion posture correction system based on drone, characterized in that: The system comprises: An exercise program clustering unit, used for collecting first physiological data of the user before exercise, clustering the first physiological data with data of the exercise guidance program case library, generating a weight training program for the user, and determining a posture guidance program according to the weight training program; A motion video acquisition unit, used to collect the second physiological data of the user during motion based on the sensor, and determine whether the user is in a normal motion mode according to the second physiological data; if not, remind the user to stop exercising; if yes, use a drone to shoot continuous frame video stream data during the user's motion; A skeleton key point extraction unit is used to divide the video stream into multiple groups of continuous frame images according to the minimum action unit, extract the corresponding user contour features based on each group of continuous frame images and calculate the fuzziness of the features, match the deblurring operation according to the fuzziness to generate the target contour features, and extract the user skeleton key points from the target contour features of the continuous frames; An angle deviation calculation unit, used to determine the actual coordinates of the user's skeleton key points in the continuous frame images and the ideal coordinates of the skeleton key points in the posture guidance scheme, and calculate the joint angle deviations of different skeleton key points of the user in the image frame motion posture according to the actual coordinates and the ideal coordinates; The motion plan optimization unit is used to calculate the target angle deviation of the user's continuous motion posture in the minimum action unit based on multiple joint angle deviations, update the user's weight training plan based on the target angle deviation, and optimize the posture guidance plan using the updated weight training plan.

5. The human motion posture correction system based on drone according to claim 4 is characterized in that: The motion scheme optimization unit is further used for: Suppose a joint consists of three key points A, B, and C. The coordinates of A, B, and C are (x A ,y A ), (x B ,y B ), (x C ,y C ), determine the vector and calculate calculate Determine θ according to the calculation formula of cosθ ij,a ,θ ij,g , then: Dth ij =|θ ij,a -θ ij,g |; In the formula, Δθ ij represents the angle deviation of the i-th joint in the j-th frame image, θ ij,a ,θ ij,g They represent the actual angle and ideal angle of the i-th joint in the j-th frame image respectively; Assuming there are m joints and n frames of images, determine the target angle deviation of the user's continuous motion posture in the minimum action unit Get the average speed of the i-th joint in n frames of images and the average acceleration In the formula, v ij 、a ij Respectively represent the velocity and acceleration of the i-th joint in the j-th frame image; To update a user's weight training program: Where W and W0 are the training weights before and after the user update, respectively; k is an adjustment factor with a value range of (0.01, 0.1); G and G0 are the user weight and the preset standard weight, respectively; and e is a natural constant.

6. The human motion posture correction system based on drone according to claim 4 is characterized in that: The motion scheme clustering unit is further used for: Acquiring first physiological data, including the user's age, height, weight, and heart rate, blood pressure, EMG signal strength, and blood oxygen concentration before exercise; Performing data cleaning and normalization operations on the first physiological data, and using the PCA algorithm to perform feature dimension reduction on the normalized data; Analyze the correlation between the features after dimensionality reduction and the load parameters, remove the features whose correlation is less than the preset threshold, and use the remaining features as the target features; The K-Means clustering algorithm is used to cluster the target features and the data features of the exercise guidance program case library to generate the user's weight training plan.

7. An electronic device, characterized in that: include: A processor and a memory, wherein the memory is used to store computer program codes, wherein the computer program codes include computer instructions, and when the processor executes the computer instructions, the electronic device executes the method for correcting human motion posture based on a drone as described in any one of claims 1 to 3.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes the method for correcting human motion posture based on a drone as described in any one of claims 1 to 3.

Images