Motion posture recognition method based on big data and image recognition

Abstract

The invention discloses a motion posture recognition method based on big data and image recognition, and the method comprises the steps: obtaining a motion video stream of a target user in real time, and extracting a user posture image sequence in the motion video stream; key point detection is carried out on the user posture image sequence, a user real-time posture skeleton diagram is generated, and a real-time multi-dimensional posture feature vector is calculated based on the skeleton diagram; based on a preset motion attitude feature big data cloud library, extracting a reference attitude feature vector set of the current motion item; matching the real-time multi-dimensional attitude feature vector with a reference attitude feature vector in a reference attitude feature vector set to obtain a matching degree; and recognizing the category and the quality grade of the current motion posture according to the matching degree. According to the method, the motion image is collected in real time, the current motion posture is compared and analyzed in combination with the motion posture feature big data cloud library, and the motion posture recognition accuracy is improved; and a basis is provided for subsequent motion posture adjustment.

Description

Technical Field

[0001] This invention relates to the field of motion posture recognition technology, and more specifically, to a motion posture recognition method based on big data and image recognition. Background Technology

[0002] With the rapid development of computer vision and artificial intelligence technologies, motion posture analysis based on image recognition has become a research hotspot in sports science, rehabilitation training, and mass fitness. Existing technologies typically capture human motion images using a single camera or depth sensor (such as Kinect), extract two-dimensional or three-dimensional human key points using open-source libraries such as OpenPose and AlphaPose, and then calculate simple parameters such as joint angles and motion trajectories. These parameters are then compared with pre-set limited standard motion models to achieve motion recognition or standardization assessment.

[0003] However, existing technologies exhibit significant limitations when applied to professional, complex, and even extreme sports scenarios. First, data silos and insufficient model generalization ability: most systems rely on standardized action datasets collected in small-scale, laboratory environments. These datasets are sparse and lack sufficient coverage of individual differences, leading to a sharp drop in recognition accuracy when faced with the ever-changing personal styles, body types, and non-standard movements in the real world. This results in an inability to provide assessments with broad reference value. Second, poor environmental adaptability: existing algorithms are mostly designed for well-lit, simple indoor environments. In complex outdoor environments (such as snow, water, and mountains), strong light, reflections, complex backgrounds, dynamic occlusion (such as water splashes and snow fog), and special clothing (such as heavy ski suits) severely interfere with image quality and key points. The system suffers from several problems: First, inaccurate detection leads to system failure. Second, the analysis is limited in scope and feedback is delayed. Technical solutions often focus on static posture or simple dynamic recognition, lacking in-depth analysis of the temporal logic of movement, the importance weight of each stage, and multi-dimensional biomechanical characteristics (such as acceleration and momentum changes). Feedback is often limited to post-event video playback and data reports, failing to provide real-time, intuitive, and actionable guidance during exercise. This delayed feedback is particularly problematic in high-speed, high-risk extreme sports, where it renders the training almost useless. Third, the system lacks personalized evolution and safety warning mechanisms. It typically adopts a one-size-fits-all standard, failing to track individual user progress and predicting and warning of potential high-risk errors based on historical data, thus lacking functionality in preventing sports injuries. Summary of the Invention

[0004] To address at least one of the aforementioned technical problems, the present invention aims to provide a motion posture recognition method based on big data and image recognition, which can improve the accuracy of motion posture recognition.

[0005] This invention provides a motion pose recognition method based on big data and image recognition, including: The motion video stream of the target user is acquired in real time, and the user posture image sequence in the motion video stream is extracted. Key point detection is performed on the user pose image sequence to generate a real-time user pose skeleton map, and a real-time multi-dimensional pose feature vector is calculated based on the skeleton map. Based on a pre-set big data cloud database of motion posture features, extract the benchmark posture feature vector set for the current sport. The matching degree is obtained by matching the real-time multi-dimensional attitude feature vector with the reference attitude feature vector set. The category and quality level of the current motion posture are identified based on the matching degree.

[0006] In this solution, the step of calculating the real-time multi-dimensional pose feature vector based on the skeleton graph specifically includes: Based on the skeleton diagram, calculate the bone vectors between the joints; Based on the bone vectors, calculate the set of angles between key bones. , where m is the number of angles of interest; Based on the changes of the skeleton map across consecutive frames in the user pose image sequence, the motion velocity and acceleration feature set of the joints is calculated. Where n is the number of joints; The angle set A and the motion feature set M are normalized and concatenated to form the real-time multi-dimensional pose feature vector. .

[0007] In this scheme, the formula for matching the real-time multi-dimensional attitude feature vector with the reference attitude feature vector set to obtain the matching degree is as follows: Where S represents the matching degree; K represents the total number of motion stages; and i belongs to k. This represents the weight coefficient for the i-th stage; This represents the real-time multi-dimensional pose feature vector at the i-th stage; This indicates that in the set of reference attitude feature vectors, and The standard pose feature vector with the closest Euclidean distance.

[0008] This plan also includes: When the matching degree is higher than the preset matching degree threshold, extract the real-time multi-dimensional pose feature vector corresponding to the current matching degree; The real-time multi-dimensional posture feature vector corresponding to the matching degree is set as the high-quality user posture feature vector and added to the baseline posture feature vector set.

[0009] This plan also includes: Set the matching degree of the current frame to The matching degree of the previous frame is set to The difference in matching degree between adjacent frames is then... ; If the difference in matching degree between adjacent frames is greater than the preset coherence factor, an alert message will be triggered. Based on the warning information, if the matching degree of the current frame is less than the preset second matching degree threshold, the preset safety liner will be triggered. Based on the warning information, if the matching degree of the current frame is greater than or equal to the preset second matching degree threshold, the attitude feature vector of the current frame is optimized according to the attitude feature vector within the preset sliding time window to recalculate the matching degree of the current frame.

[0010] This plan also includes: Once the warning message is triggered, the current video frame location is marked. After traversing all motion video streams, obtain all marked positions; Using any marked location as the center and a distance as the radius, construct an influence range and extract the number of times the mark is made within the influence range; If the number of markings within the affected area exceeds a preset threshold, a maintenance warning message for the current affected area will be generated.

[0011] This plan also includes: Based on a preset sliding time window, obtain the matching degree set for different time periods; The average matching score for the corresponding time period is obtained by averaging the matching scores in the matching score set. Subtract the average matching score of the previous time period from the average matching score of the current time period to obtain the average matching score difference. If the absolute value of the average matching degree difference is less than the preset first average matching degree threshold, a smooth motion state prompt message is generated. If the average matching degree difference is greater than the preset second average matching degree threshold, a motion state improvement prompt message will be generated. If the average matching degree difference is less than the preset third average matching degree threshold, a motion state decline prompt message will be generated.

[0012] In this solution, after generating the motion state decline warning information, the solution further includes: Extract the matching score set for the current time period, and extract the minimum matching score in the corresponding matching score set; The real-time multi-dimensional pose feature vector for the current motion is determined based on the minimum matching degree in the corresponding matching degree set. The difference between the real-time multi-dimensional posture feature vector during the current motion and the corresponding standard posture feature vector is calculated to obtain the motion posture feature vector adjustment value. The adjusted values ​​of the motion posture feature vector are sent to a preset screen for display.

[0013] In this solution, after performing key point detection on the user pose image sequence to generate a real-time user pose skeleton map, the solution further includes: If the current activity is skiing, the user's body surface temperature distribution map will be obtained in real time based on the preset infrared thermal imaging sensor. Cluster analysis is performed on the temperature point cloud in the user's body surface temperature distribution map to obtain the point cloud map of the user's joints and determine the range of the user's joints. If the joints in the user's real-time pose skeleton graph fall within the corresponding joint range, then the current user's real-time pose skeleton graph is normal. If a joint in the user's real-time pose skeleton graph does not fall within the corresponding joint range, the position of that joint will be revised based on the corresponding joint range.

[0014] One or more technical solutions proposed in this application have at least the following technical effects: By building and continuously updating a big data cloud library of motion posture features, the set of benchmark posture feature vectors is constantly enriched and evolved, thereby improving the accuracy and professionalism of structure identification and evaluation. For the special scenario of extreme skiing on snow, an infrared thermal imaging sensor is added to obtain the user's body surface temperature distribution map. By taking advantage of the high temperature at the joint points, the range of the user's joint points can be identified. Combined with the user's real-time posture skeleton map, the error in real-time posture skeleton recognition caused by the user wearing heavy clothing while skiing on snow is reduced. In summary, this invention, by constructing a big data cloud database of motion posture features and combining it with deep image processing and intelligent analysis algorithms for specific scenarios, achieves a comprehensive breakthrough in motion posture recognition technology in terms of accuracy, reliability, real-time performance, security, and personalization. It provides a complete, efficient, and intelligent solution for high-performance sports training, public scientific fitness, and rehabilitation assessment. Attached Figure Description

[0015] Figure 1 The flowchart of the motion posture recognition method based on big data and image recognition of the present invention is shown. Detailed Implementation

[0016] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0017] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0018] Figure 1 The flowchart of the motion posture recognition method based on big data and image recognition of the present invention is shown.

[0019] like Figure 1 As shown, this invention discloses a motion posture recognition method based on big data and image recognition, including: S101, acquire the motion video stream of the target user in real time, and extract the user posture image sequence from the motion video stream; S102, perform key point detection on the user pose image sequence to generate a real-time user pose skeleton map, and calculate a real-time multi-dimensional pose feature vector based on the skeleton map. S103, based on a preset big data cloud library of motion posture features, extracts the benchmark posture feature vector set of the current sport; S104, Match the real-time multi-dimensional attitude feature vector with the reference attitude feature vector in the reference attitude feature vector set to obtain the matching degree; S105 identifies the category and quality level of the current motion posture based on the matching degree.

[0020] According to an embodiment of the present invention, raw data is acquired through image acquisition, and digital feature vectors representing the essence of user posture are extracted through key point detection and skeleton graph generation. The preset motion posture feature big data cloud database contains a set of baseline posture feature vectors for various sports, such as table tennis. The baseline posture feature vector corresponding to the current multi-dimensional posture feature vector is determined based on the matching degree, and the current motion posture category is determined based on the baseline posture feature vector. Each baseline posture feature vector in the set of baseline posture feature vectors corresponds to a motion posture category, such as "giant slalom" in skiing. The quality of the corresponding motion posture category is then determined by the matching degree. The matching degree is divided into multiple matching degree ranges, and each matching degree range corresponds to a quality level. The higher the matching degree in the matching degree range, the higher the corresponding quality level.

[0021] According to an embodiment of the present invention, the step of calculating a real-time multi-dimensional pose feature vector based on the skeleton graph specifically includes: Based on the skeleton diagram, calculate the bone vectors between the joints; Based on the bone vectors, calculate the set of angles between key bones. , where m is the number of angles of interest; Based on the changes of the skeleton map across consecutive frames in the user pose image sequence, the motion velocity and acceleration feature set of the joints is calculated. Where n is the number of joints; The angle set A and the motion feature set M are normalized and concatenated to form the real-time multi-dimensional pose feature vector. .

[0022] It's important to note that the set of angles A between skeletal vectors and key bones (e.g., knee angle, torso lean angle) precisely describes the relative spatial relationships of various body parts at a given instant in a frame. This is the core basis for assessing whether a posture is "in place." By calculating the set of velocities and accelerations M at joint points, the rate and rhythm of posture changes are captured. This is crucial for assessing the smoothness and explosiveness of movements, as well as identifying specific dynamic errors (such as excessive deceleration). For example, in turning, the angular acceleration of the hip joint effectively reflects the agility of turning force. Normalizing (to eliminate the influence of individual body shape differences) and concatenating the static A and dynamic M forms a unified, high-information-density mathematical expression. .

[0023] According to an embodiment of the present invention, the formula for matching the real-time multi-dimensional attitude feature vector with the reference attitude feature vector set to obtain the matching degree is as follows: Where S represents the matching degree; K represents the total number of motion stages; and i belongs to k. This represents the weight coefficient for the i-th stage; This represents the real-time multi-dimensional pose feature vector at the i-th stage; This indicates that in the set of reference attitude feature vectors, and The standard pose feature vector with the closest Euclidean distance.

[0024] It should be noted that the action cycle is decomposed into K stages for segmented matching, and different weight coefficients are set for different stages, such as assigning higher weight coefficients to key stages. The total score S is obtained by calculating the normalized distance between the user's feature vector at each stage and the nearest standard vector in the cloud library, and then summing them with weights. This score intuitively reflects the overall similarity between the user's action and the ideal action. For example, S=0.85 can be set as good action completion quality. D represents the normalized distance function, which is used to calculate the difference between two corresponding feature vectors. The smaller the D value, the more similar they are.

[0025] According to an embodiment of the present invention, it further includes: When the matching degree is higher than the preset matching degree threshold, extract the real-time multi-dimensional pose feature vector corresponding to the current matching degree; The real-time multi-dimensional posture feature vector corresponding to the matching degree is set as the high-quality user posture feature vector and added to the baseline posture feature vector set.

[0026] It should be noted that with the use of a massive number of users, the big data cloud library can collect action patterns of different body types and styles that are equally efficient, thereby making the standard set more diverse and inclusive. It can even discover and summarize better action paradigms. For example, if the preset matching degree threshold is 0.85, the real-time multi-dimensional posture feature vectors corresponding to matching degrees higher than 0.85 will be fed back to the corresponding baseline posture feature vector set for storage.

[0027] Furthermore, the reference attitude feature vectors in the reference attitude feature vector set are sorted according to the order of their storage time. When the number of reference attitude feature vectors exceeds a preset threshold, the earliest reference attitude feature vector is deleted until the number of reference attitude feature vectors in the reference attitude feature vector set does not exceed the preset threshold.

[0028] According to an embodiment of the present invention, it further includes: Set the matching degree of the current frame to The matching degree of the previous frame is set to The difference in matching degree between adjacent frames is then... ; If the difference in matching degree between adjacent frames is greater than the preset coherence factor, an alert message will be triggered. Based on the warning information, if the matching degree of the current frame is less than the preset second matching degree threshold, the preset safety liner will be triggered. Based on the warning information, if the matching degree of the current frame is greater than or equal to the preset second matching degree threshold, the attitude feature vector of the current frame is optimized according to the attitude feature vector within the preset sliding time window to recalculate the matching degree of the current frame.

[0029] It should be noted that in normal continuous motion, the posture matching degree between adjacent frames should change smoothly. When the difference in matching degree between adjacent frames is greater than the preset coherence factor, the drastic difference in the current frame is determined to be an anomaly caused by external interference. If the matching degree of the current frame is less than the preset second matching degree threshold, it indicates that the posture feature vector of the current motion differs significantly from the standard posture feature vector, which may easily cause injury to the athlete, thus triggering the preset safety cushion. If the matching degree of the current frame is greater than or equal to the preset second matching degree threshold, it indicates that the posture feature vector of the current motion differs slightly from the standard posture feature vector, the motion posture is relatively normal, and no injury will be caused. Further, the historical feature vectors of the current frame and its N previous frames are extracted and subjected to weighted averaging or median filtering to obtain the posture feature vector after optimization in the current frame. The preset second matching degree threshold is less than the preset matching degree threshold.

[0030] According to an embodiment of the present invention, it further includes: Once the warning message is triggered, the current video frame location is marked. After traversing all motion video streams, obtain all marked positions; Using any marked location as the center and a distance as the radius, construct an influence range and extract the number of times the mark is made within the influence range; If the number of markings within the affected area exceeds a preset threshold, a maintenance warning message for the current affected area will be generated.

[0031] It should be noted that if multiple video streams are marked at the same location, and the number of marks exceeds a preset threshold (e.g., 3), it indicates that the current location may be located in an unsuitable sports field, such as having too small a curve or being uneven. Therefore, a warning message is generated to prompt timely action.

[0032] According to an embodiment of the present invention, it further includes: Based on a preset sliding time window, obtain the matching degree set for different time periods; The average matching score for the corresponding time period is obtained by averaging the matching scores in the matching score set. Subtract the average matching score of the previous time period from the average matching score of the current time period to obtain the average matching score difference. If the absolute value of the average matching degree difference is less than the preset first average matching degree threshold, a smooth motion state prompt message is generated. If the average matching degree difference is greater than the preset second average matching degree threshold, a motion state improvement prompt message will be generated. If the average matching degree difference is less than the preset third average matching degree threshold, a motion state decline prompt message will be generated.

[0033] It should be noted that the preset third average matching degree threshold is less than zero and the absolute value of the preset third average matching degree threshold is greater than the preset first average matching degree threshold. The preset first average matching degree threshold is greater than zero and less than the preset second average matching degree threshold. Through a preset sliding time window, the matching degree set of the athlete within a certain period of time is extracted and compared with the matching degree set of adjacent time periods. The average matching degree of the corresponding time period is used to quantitatively assess whether the athlete has reached or is close to the personal peak level in the current state and monitor the fluctuations in state. For example, if the difference in average matching degree is greater than the preset second average matching degree threshold, a sports state improvement prompt message such as "Your turning angle stability today has improved by 5% compared with the best record in the previous stage" is generated. This personalized feedback can greatly improve user stickiness and training motivation, transforming the system from a cold action referee into a personalized intelligent coach that understands the user and accompanies the user's growth.

[0034] According to an embodiment of the present invention, after generating the motion state decline prompt information, the method further includes: Extract the matching score set for the current time period, and extract the minimum matching score in the corresponding matching score set; The real-time multi-dimensional pose feature vector for the current motion is determined based on the minimum matching degree in the corresponding matching degree set. The difference between the real-time multi-dimensional posture feature vector during the current motion and the corresponding standard posture feature vector is calculated to obtain the motion posture feature vector adjustment value. The adjusted values ​​of the motion posture feature vector are sent to a preset screen for display.

[0035] It should be noted that after generating the motion state decline prompt information, by finding the real-time multi-dimensional posture vector corresponding to the minimum matching degree in the current time period, and by making targeted adjustments to the current motion posture feature vector based on the feature difference between the real-time multi-dimensional posture vector and the corresponding standard posture feature vector, the rationality of the motion can be effectively improved.

[0036] According to an embodiment of the present invention, after performing key point detection on the user pose image sequence to generate a real-time user pose skeleton map, the method further includes: If the current activity is skiing, the user's body surface temperature distribution map will be obtained in real time based on the preset infrared thermal imaging sensor. Cluster analysis is performed on the temperature point cloud in the user's body surface temperature distribution map to obtain the point cloud map of the user's joints and determine the range of the user's joints. If the joints in the user's real-time pose skeleton graph fall within the corresponding joint range, then the current user's real-time pose skeleton graph is normal. If a joint in the user's real-time pose skeleton graph does not fall within the corresponding joint range, the position of that joint will be revised based on the corresponding joint range.

[0037] It should be noted that if the current sports scenario is skiing, it means that the athlete is wearing heavy clothing, and there may be errors in recognizing the user's real-time posture skeleton map through video stream. Therefore, by combining the user's body surface temperature distribution map for comprehensive analysis, the accuracy of the joint position in the user's real-time posture skeleton map can be improved.

[0038] Furthermore, if a joint in the user's real-time posture skeleton diagram does not fall within the corresponding joint range, when time has passed after the joint range, the point closest to the corresponding joint in the joint range is extracted and set as the revised position of the joint; when time has passed before the joint range, the point closest to the corresponding joint in the joint range is extracted and set as the nearest point. The position of the nearest point is connected to the position of the joint in the user's real-time posture skeleton diagram. Based on a preset weight, the position on the corresponding line is extracted and set as the revised position of the joint. For example, if the preset weight is set to 0.5, then the midpoint of the corresponding line is taken as the revised joint position.

[0039] According to an embodiment of the present invention, it further includes: If the current motion scenario is skiing, then extract the baseline posture feature vector subset based on the current ski slope gradient and snow hardness; The standard postures of the baseline posture feature vector set in the preset motion posture feature big data cloud library have been labeled with their applicable slope range and snow hardness range. Based on the real-time slope information fed back by the slope sensor and the snow hardness selected by the user, the reference attitude feature vector set of the current environment is dynamically selected, so that the reference attitude feature vector is adaptively adjusted with environmental conditions.

[0040] This invention discloses a motion posture recognition method based on big data and image recognition, comprising: acquiring a motion video stream of a target user in real time and extracting a sequence of user posture images from the motion video stream; performing key point detection on the user posture image sequence to generate a real-time user posture skeleton map, and calculating a real-time multi-dimensional posture feature vector based on the skeleton map; extracting a baseline posture feature vector set for the current sport based on a preset motion posture feature big data cloud library; matching the real-time multi-dimensional posture feature vector with the baseline posture feature vector set to obtain a matching degree; and identifying the category and quality level of the current motion posture based on the matching degree. This invention improves the accuracy of motion posture recognition by acquiring motion images in real time and combining them with a motion posture feature big data cloud library for comparative analysis of the current motion posture; and provides a basis for subsequent motion posture adjustments.

[0041] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0042] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0043] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0044] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0045] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

Claims

1. A motion posture recognition method based on big data and image recognition, characterized in that, include: The motion video stream of the target user is acquired in real time, and the user posture image sequence in the motion video stream is extracted. Key point detection is performed on the user pose image sequence to generate a real-time user pose skeleton map, and a real-time multi-dimensional pose feature vector is calculated based on the skeleton map. Based on a pre-set big data cloud database of motion posture features, extract the benchmark posture feature vector set for the current sport. The matching degree is obtained by matching the real-time multi-dimensional attitude feature vector with the reference attitude feature vector set. The category and quality level of the current motion posture are identified based on the matching degree.

2. The motion posture recognition method based on big data and image recognition according to claim 1, characterized in that, The steps for calculating real-time multi-dimensional pose feature vectors based on the skeleton graph specifically include: Based on the skeleton diagram, calculate the bone vectors between the joints; Based on the bone vectors, calculate the set of angles between key bones. , where m is the number of angles of interest; Based on the changes of the skeleton map across consecutive frames in the user pose image sequence, the motion velocity and acceleration feature set of the joints is calculated. Where n is the number of joints; The angle set A and the motion feature set M are normalized and concatenated to form the real-time multi-dimensional pose feature vector. .

3. The motion posture recognition method based on big data and image recognition according to claim 1, characterized in that, The formula for matching the real-time multi-dimensional attitude feature vector with the reference attitude feature vector set to obtain the matching degree is as follows: Where S represents the matching degree; K represents the total number of motion stages; and i belongs to k. This represents the weight coefficient for the i-th stage; This represents the real-time multi-dimensional pose feature vector at the i-th stage; This indicates that in the set of reference attitude feature vectors, and The standard pose feature vector with the closest Euclidean distance.

4. The motion posture recognition method based on big data and image recognition according to claim 3, characterized in that, Also includes: When the matching degree is higher than the preset matching degree threshold, extract the real-time multi-dimensional pose feature vector corresponding to the current matching degree; The real-time multi-dimensional posture feature vector corresponding to the matching degree is set as the high-quality user posture feature vector and added to the baseline posture feature vector set.

5. The motion posture recognition method based on big data and image recognition according to claim 1, characterized in that, Also includes: Set the matching degree of the current frame to The matching degree of the previous frame is set to The difference in matching degree between adjacent frames is then... ; If the difference in matching degree between adjacent frames is greater than the preset coherence factor, an alert message will be triggered. Based on the warning information, if the matching degree of the current frame is less than the preset second matching degree threshold, the preset safety liner will be triggered. Based on the warning information, if the matching degree of the current frame is greater than or equal to the preset second matching degree threshold, the attitude feature vector of the current frame is optimized according to the attitude feature vector within the preset sliding time window to recalculate the matching degree of the current frame.

6. The motion posture recognition method based on big data and image recognition according to claim 5, characterized in that, Also includes: Once the warning message is triggered, the current video frame location is marked. After traversing all motion video streams, obtain all marked positions; Using any marked location as the center and a distance as the radius, construct an influence range and extract the number of times the mark is made within the influence range; If the number of markings within the affected area exceeds a preset threshold, a maintenance warning message for the current affected area will be generated.

7. The motion posture recognition method based on big data and image recognition according to claim 5, characterized in that, Also includes: Based on a preset sliding time window, obtain the matching degree set for different time periods; The average matching score for the corresponding time period is obtained by averaging the matching scores in the matching score set. Subtract the average matching score of the previous time period from the average matching score of the current time period to obtain the average matching score difference. If the absolute value of the average matching degree difference is less than the preset first average matching degree threshold, a smooth motion state prompt message is generated. If the average matching degree difference is greater than the preset second average matching degree threshold, a motion state improvement prompt message will be generated. If the average matching degree difference is less than the preset third average matching degree threshold, a motion state decline prompt message will be generated.

8. The motion posture recognition method based on big data and image recognition according to claim 7, characterized in that, After generating the motion state decline warning message, the method further includes: Extract the matching score set for the current time period, and extract the minimum matching score in the corresponding matching score set; The real-time multi-dimensional pose feature vector for the current motion is determined based on the minimum matching degree in the corresponding matching degree set. The difference between the real-time multi-dimensional posture feature vector during the current motion and the corresponding standard posture feature vector is calculated to obtain the motion posture feature vector adjustment value. The adjusted values ​​of the motion posture feature vector are sent to a preset screen for display.

9. The motion posture recognition method based on big data and image recognition according to claim 1, characterized in that, After performing key point detection on the user pose image sequence to generate a real-time user pose skeleton map, the method further includes: If the current activity is skiing, the user's body surface temperature distribution map will be obtained in real time based on the preset infrared thermal imaging sensor. Cluster analysis is performed on the temperature point cloud in the user's body surface temperature distribution map to obtain the point cloud map of the user's joints and determine the range of the user's joints. If the joints in the user's real-time pose skeleton graph fall within the corresponding joint range, then the current user's real-time pose skeleton graph is normal. If a joint in the user's real-time pose skeleton graph does not fall within the corresponding joint range, the position of that joint will be revised based on the corresponding joint range.

Images