Personalized follow-up training guidance method and device, storage medium and program product

Abstract

This disclosure provides a personalized follow-up training guidance method, device, storage medium, and program product. The method includes: real-time acquisition of a user's follow-up training video; real-time posture evaluation of the video to obtain a dynamic skeletal diagram; multi-dimensional evaluation of the user's movements based on the skeletal diagram and a standard action video library to obtain a score for each evaluation dimension; identification of the user's error type based on the score for each evaluation dimension; determination and real-time feedback of a suitable guidance strategy to the user based on a preset user profile and the error type; evaluation of the guidance effect based on the score and error type; and updating the user profile based on the guidance effect. This method enables personalized adaptive guidance throughout the entire process of exercise follow-up training through multi-dimensional action evaluation, accurate error identification, and dynamic profile updates, effectively improving movement standardization and training safety.

Description

Technical Field

[0001] This disclosure relates to the field of computer vision technology, and in particular to a personalized follow-up guidance method, device, storage medium, and program product. Background Technology

[0002] Currently, video-based home fitness apps are becoming increasingly popular due to their convenience and flexibility, serving as an important tool for public health management. Existing technologies mainly fall into two categories: one is the pre-recorded standard video follow-up mode used by most apps, which guides users through training by playing instructional content in a one-way manner; the other is a few solutions with basic movement assessment functions, which can collect user movement data based on ordinary cameras or wearable sensors and attempt to perform simple analysis to provide feedback.

[0003] However, pre-recorded videos cannot recognize and assess users' actual movements in real time, making it difficult for users to judge the standard of their movements. This not only reduces the effectiveness of exercise but also increases the risk of sports injuries. Applications with assessment functions often rely on simple joint angle thresholds, which are ill-suited to varying user fitness levels and flexibility. Even when errors are detected, the system typically only provides general feedback (such as "movement is not standard"), failing to diagnose the specific cause of the error or provide actionable, personalized improvement suggestions, thus limiting the effectiveness of guidance and the user experience. Summary of the Invention

[0004] In view of this, the present disclosure provides a personalized training guidance method, device, storage medium, and program product, which can realize personalized adaptive guidance throughout the entire process of sports training through multi-dimensional motion evaluation, accurate error identification, and dynamic profile updates, effectively improving the standardization of movements and training safety.

[0005] Firstly, this disclosure provides a personalized follow-up guidance method, employing the following technical solution: The system collects users' practice videos in real time, performs real-time posture evaluation on the practice videos, and obtains dynamic skeletal diagrams. Based on the skeletal diagram and standard motion video library, the user's actions are evaluated in multiple dimensions, and a score value is obtained for each evaluation dimension. Based on the score value of each evaluation dimension, identify the user's error type; Based on the preset user profile and the error type, determine and provide appropriate guidance strategies to the user in real time; The effectiveness of the guidance is evaluated based on the rating and the error type, and the user profile is updated based on the effectiveness of the guidance.

[0006] Optionally, based on the skeletal diagram and standard motion video library, the user's movements are evaluated in multiple dimensions to obtain a score for each evaluation dimension, including: Extract target standard action videos from the standard action video library, segment the target standard action videos according to action stages, and obtain multiple action stage sub-videos; Obtain the baseline threshold range for each action feature in each action stage sub-video across multiple evaluation dimensions; Based on the user profile, the baseline threshold range is dynamically adjusted to obtain a personalized threshold range that suits the user. Based on multiple evaluation dimensions, the user's action feature values ​​are extracted in real time from the dynamic skeleton map of the current action stage; Match the action feature values ​​of each evaluation dimension in the current action phase with the corresponding personalized threshold range; Based on the matching results, obtain the score value for each action feature under each evaluation dimension.

[0007] Optionally, the step of dynamically adjusting the baseline threshold range based on the user profile to obtain a personalized threshold range suitable for the user includes: Based on the user profile, the user's physical fitness level and physical condition information are obtained; based on the training video, the user's real-time status data is obtained. Based on the aforementioned physical condition information, the health level of each key limb component of the user is obtained; An overall basic adjustment coefficient is set based on the aforementioned physical fitness level. Based on the health level of each key limb component, local feature constraint coefficients are set for each movement feature under multiple evaluation dimensions; A temporary elastic correction coefficient is set based on the real-time status data; Based on the overall basic adjustment coefficient, the local feature constraint coefficient, and the temporary elastic correction coefficient, the benchmark threshold range is dynamically adjusted to obtain a personalized threshold range suitable for the user.

[0008] Optionally, the evaluation dimensions include at least one of posture, temporality, and stability; the step of extracting the user's action feature values ​​from the dynamic skeleton map of the current action stage in real time according to multiple evaluation dimensions includes: Real-time analysis of the dynamic skeleton map corresponding to the current action stage to determine the target joints and core stability key points required to complete the current target action; Acquire spatial angle data of the joints used by the target, temporal data of continuous frame motion changes, and real-time coordinate offset data of the core stabilization key points within a preset time period; The spatial angle data of the joints of each target are used as the motion feature values ​​of the corresponding limb posture features under the posture evaluation dimension; Based on the continuous frame motion change time-series data of the joints used by each target, the motion feature values ​​of the corresponding limb motion rhythm features under the time-series evaluation dimension are obtained. Based on the coordinate offset data of the core stability key points, the motion feature values ​​of the corresponding limb sway control features under the stability assessment dimension are obtained.

[0009] Optionally, obtaining the score value for each action feature under each evaluation dimension based on the matching results includes: When the action feature value is within the corresponding personalized threshold range, the corresponding score value is set to the preset maximum value; When the action feature value is not within the corresponding personalized threshold range, find the boundary value that is closest to the action feature value from the personalized threshold range; The corresponding score is obtained based on the difference between the nearest boundary value and the action feature value.

[0010] Optionally, determining and providing appropriate guidance strategies to the user in real time based on the preset user profile and the error type includes: Extract user profile parameters from the user profile; Based on the user profile parameters, the error type, and the corresponding score, construct the query conditions; Based on the query conditions, a suitable guidance strategy is matched from the guidance strategy rule base.

[0011] Optionally, updating the user profile based on the guidance effect includes: Based on the guidance effect, update the user's physical fitness level and physical condition information; The user profile is updated based on the new fitness level and physical condition information.

[0012] Secondly, this disclosure also provides a personalized follow-up guidance system, which adopts the following technical solution: The real-time posture assessment module is used to collect the user's follow-up video in real time, perform real-time posture assessment on the follow-up video, and obtain a dynamic skeleton map. The user action evaluation module is used to evaluate the user's actions in multiple dimensions based on the skeletal diagram and standard action video library, and obtain a score value for each evaluation dimension. The error type identification module is used to identify the user's error type based on the score value of each evaluation dimension; The guidance strategy determination module is used to determine and provide appropriate guidance strategies to the user in real time based on the preset user profile and the error type. The user profile update module is used to evaluate the guidance effect based on the rating value and the error type, and update the user profile based on the guidance effect.

[0013] Thirdly, this disclosure also provides a computer device, which adopts the following technical solution: The computer device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform any of the personalized follow-up guidance methods described above.

[0014] Fourthly, embodiments of this disclosure also provide a computer-readable storage medium storing computer instructions for causing a computer to execute any of the personalized follow-up guidance methods described above.

[0015] Fifthly, embodiments of this disclosure also provide a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of any of the methods described above.

[0016] The personalized follow-up training guidance method provided in this disclosure generates a dynamic skeletal map by real-time acquisition of user follow-up training videos and real-time posture assessment. Based on ordinary visual acquisition methods, it can achieve refined and real-time perception of the human joint and limb movement status without the need for additional professional sensing equipment, balancing ease of acquisition and real-time performance, and providing reliable underlying data support for subsequent accurate movement analysis. Combining the dynamic skeletal map with a pre-built standard movement video library to conduct multi-dimensional movement assessment and output score values ​​for each dimension, it can overcome the limitations of traditional single-indicator judgment, quantitatively representing the user's movement completion from multiple dimensions, and achieving a comprehensive and objective quantitative assessment of movement quality. By using the score values ​​of each dimension to identify specific movement error types, it can accurately locate the specific form and cause of movement deviation problems, avoiding generalized movement judgments and achieving refined movement problem diagnosis. By combining pre-set user profiles with identified error types and providing real-time feedback on adapted guidance strategies, customized error correction guidance can be output based on individual user attributes such as fitness levels and physical conditions. This overcomes the limitations of uniform guidance, improves the targeting and effectiveness of movement correction, and optimizes the real-time training experience. Furthermore, the actual implementation effect of guidance is comprehensively evaluated using scoring values ​​and error types, and user profiles are dynamically updated based on the guidance results. This allows for continuous adaptation to long-term individual differences such as improved athletic ability and changes in status, forming a closed-loop mechanism of movement assessment, error diagnosis, personalized guidance, and dynamic profile optimization. This continuously optimizes the adaptability of subsequent movement assessment standards and guidance strategies, effectively standardizes user training movements, reduces the risk of sports injuries caused by incorrect training, and improves training quality and user engagement in the long term.

[0017] The above description is merely an overview of the technical solution disclosed herein. In order to better understand the technical means of this disclosure and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this disclosure more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart illustrating the personalized follow-up guidance method provided in this embodiment of the disclosure; Figure 2 A flowchart illustrating the multi-dimensional evaluation method provided in the embodiments of this disclosure; Figure 3A flowchart illustrating the method for dynamically adjusting the reference threshold range provided in this embodiment of the disclosure; Figure 4 A flowchart illustrating the action feature value acquisition method provided in this embodiment of the disclosure; Figure 5 A flowchart illustrating the scoring value acquisition method provided in this embodiment of the disclosure; Figure 6 A flowchart illustrating the guidance strategy acquisition method provided in this embodiment of the disclosure; Figure 7 A flowchart illustrating the user profile update method provided in this embodiment of the disclosure; Figure 8 A schematic diagram of the personalized follow-up guidance system provided in the embodiments of this disclosure; Figure 9 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present disclosure. Detailed Implementation

[0020] The embodiments of this disclosure will now be described in detail with reference to the accompanying drawings.

[0021] It should be understood that the following specific examples illustrate the implementation of this disclosure, and those skilled in the art can easily understand other advantages and effects of this disclosure from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. This disclosure can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this disclosure. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0022] It should be noted that various aspects of embodiments within the scope of the appended claims are described below. It will be apparent that the aspects described herein can be embodied in a wide variety of forms, and any particular structure and / or function described herein is merely illustrative. Based on this disclosure, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects set forth herein can be used to implement the device and / or practice the method. Additionally, this device and / or method can be implemented using structures and / or functionalities other than one or more of the aspects set forth herein.

[0023] It should also be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this disclosure. The drawings only show the components related to this disclosure and are not drawn according to the number, shape and size of the components in actual implementation. In actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0024] Furthermore, specific details are provided in the following description to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that the described aspects can be practiced without these specific details.

[0025] Reference Figure 1 This disclosure provides a personalized follow-up guidance method, including the following steps: S1: Real-time acquisition of users' follow-up videos, real-time posture evaluation of the follow-up videos, and acquisition of dynamic skeleton maps; S2: Based on skeletal diagrams and a standard motion video library, the user's actions are evaluated in multiple dimensions, and a score value is obtained for each evaluation dimension; S3: Identify the user's error type based on the score value of each evaluation dimension; S4: Based on preset user profiles and error types, determine and provide appropriate guidance strategies to users in real time; S5: Evaluate the effectiveness of guidance based on ratings and error types, and update user profiles based on the effectiveness of guidance.

[0026] The personalized follow-up training guidance method disclosed herein generates a dynamic skeletal map by real-time acquisition of user follow-up training videos and real-time posture assessment. Based on ordinary visual acquisition methods, it achieves refined and real-time perception of the human joint and limb movement status without the need for additional professional sensing equipment, balancing ease of acquisition and real-time performance, and providing reliable underlying data support for subsequent accurate movement analysis. By combining the dynamic skeletal map with a pre-built standard movement video library to conduct multi-dimensional movement assessments and output scores for each dimension, it overcomes the limitations of traditional single-indicator judgments, quantitatively representing the user's movement completion from multiple dimensions, and achieving a comprehensive and objective quantitative assessment of movement quality.

[0027] By using multi-dimensional scoring values ​​to identify specific movement error types, the system can accurately pinpoint the specific form and cause of movement deviations, avoiding generalized movement judgments and achieving refined movement problem diagnosis. Combining pre-set user profiles with identified error types and providing real-time feedback on adapted guidance strategies, the system can output customized error correction guidance based on individual user attributes such as fitness levels and physical conditions. This overcomes the limitations of uniform guidance, improving the targetedness and effectiveness of movement correction, and optimizing the real-time training experience. Furthermore, by comprehensively evaluating the actual implementation effect of guidance using scoring values ​​and error types, and dynamically updating user profiles based on the guidance results, the system can continuously adapt to long-term individual differences such as user performance improvement and status changes. This forms a closed-loop mechanism of movement assessment, error diagnosis, personalized guidance, and dynamic profile optimization, continuously optimizing the adaptability of subsequent movement assessment standards and guidance strategies. This effectively standardizes user training movements, reduces the risk of sports injuries caused by incorrect training, and improves training quality and user engagement in the long term.

[0028] In S1, the user's training video stream is collected in real time through the user terminal device's camera, and continuous video frames are continuously output. The position of the core joints of the human body, such as the shoulder, elbow, hip, and knee, is detected frame by frame according to the lightweight real-time posture evaluation model. The coordinates of the key points of the skeleton in a single frame are extracted and a single frame skeleton map is generated. The key point data of the continuous video frames are linked together to form a time-series dynamic skeleton key point sequence, thereby constructing a user dynamic skeleton map that is updated synchronously throughout the process.

[0029] In S2, the evaluation dimensions include at least one of posture, timing, and stability. Standard movement videos of various fitness exercises are collected in batches as basic materials. Each standard movement video is broken down into multiple movement stages based on a complete, independent movement unit. For example, 20 sets of leg kicks are independent movement stages. For each independent movement stage, massive amounts of standard movement sample data are collected and statistically analyzed from the three evaluation dimensions of posture, timing, and stability to obtain the distribution ranges of various movement features such as key joint angles, movement rhythm and speed, and body sway. Based on the distribution ranges, the baseline threshold ranges for each movement feature under each evaluation dimension are statistically determined. Finally, using movement type as the index and movement stage as the sub-unit, the baseline threshold ranges for each movement feature of each standard movement video under multiple evaluation dimensions are structurally stored, forming a comprehensive standard movement video library covering multiple categories of exercises. When users follow any exercise, they can quickly retrieve and match the corresponding target standard movement video and its accompanying standard data.

[0030] Reference Figure 2 The flowchart illustrating the multi-dimensional evaluation method, which "assesses user movements from multiple dimensions based on skeletal diagrams and a standard motion video library, and obtains a score for each evaluation dimension," includes the following steps: S21: Extract the target standard motion video from the standard motion video library, segment the target standard motion video according to the motion stage, and obtain multiple motion stage sub-videos; S22: Obtain the baseline threshold range for each action feature in each action stage sub-video across multiple evaluation dimensions; S23: Based on user profiles, dynamically adjust the baseline threshold range to obtain a personalized threshold range that suits the user. S24: Extract the user's action feature values ​​from the dynamic skeleton map of the current action stage in real time according to multiple evaluation dimensions; S25: Match the action feature value of each evaluation dimension in the current action phase with the corresponding personalized threshold range; S26: Based on the matching results, obtain the score value of each action feature under each evaluation dimension.

[0031] In S21 and S22, after the user selects a specific standard movement video under the fitness exercise program to follow along with, the video is immediately extracted from the standard movement video library as the target standard movement video. At the same time, according to the pre-divided independent movement units in the library (such as a set of kicks, a set of squats), the target standard movement video is divided into multiple sub-videos corresponding to the movement stages. Simultaneously, the baseline threshold range of various movement features of each sub-video in the three evaluation dimensions of posture, timing, and stability is loaded.

[0032] In S23, refer to Figure 3 The flowchart illustrating the method for dynamically adjusting the baseline threshold range, which involves "dynamically adjusting the baseline threshold range based on user profiles to obtain a personalized threshold range suitable for the user," includes the following steps: S231: Based on user profiles, obtain users' physical fitness level and physical condition information; based on training videos, obtain users' real-time status data. S232: Based on physical condition information, obtain the health level of each key limb component of the user; S233: Set the overall basic adjustment coefficient based on the physical fitness level; S234: Set local feature constraint coefficients for each motion feature under multiple assessment dimensions based on the health level of each key limb component; S235: Set a temporary elastic correction coefficient based on real-time status data; S236: Based on the overall basic adjustment coefficient, local feature constraint coefficient and temporary elastic correction coefficient, the benchmark threshold range is dynamically adjusted to obtain a personalized threshold range suitable for the user.

[0033] In S231, a structured user profile is pre-built. This profile is a static file exclusive to each user, mainly composed of basic physical data, basic exercise information, and historical long-term training behavior data submitted by the user during registration. The physical fitness level is initially determined based on the user's initial questionnaire and basic movement assessment results, divided into multiple levels such as beginner and advanced, and stored within the user profile. Physical condition information is simultaneously included in the user profile, specifically including key joint injury history, limb flexibility, joint movement limitations, local muscle strength, postural defects, chronic sports injuries, and contraindicated limb areas. Simultaneously, the system analyzes continuous frame skeletal changes and movement degradation trends in the user's current training video in real time, independently collecting information such as training duration, frequency of movement deformation, postural deviation fluctuations, and the degree of movement rate decay, generating real-time status data independent of the user profile.

[0034] In S232, based on the physical condition information within the user profile, key limb components such as the knee, hip, shoulder, elbow, lower back, and ankle are uniformly defined and classified into three levels: weak, normal, and excellent, according to a unified standard. If the corresponding component has a history of injury, limited mobility, or a record of chronic injury, it is judged as weak; if there is no record of injury and the range of motion and flexibility meet the average standard, it is judged as normal; if there is no injury and the flexibility and muscle control are better than average, it is judged as excellent, thus achieving standardized judgment of the health level of each limb component.

[0035] In S233, a fixed range of overall basic adjustment coefficients are matched according to the physical fitness level. For example, the beginner level corresponds to a coefficient greater than 1, which is used to widen the tolerance range of the benchmark threshold; the regular level uses a coefficient of 1 to keep the benchmark threshold unchanged; the advanced level uses a coefficient less than 1 to narrow the threshold range and increase the assessment rigor. This coefficient is a globally unified coefficient that uniformly applies to the action feature thresholds of all assessment dimensions of posture, timing, and stability.

[0036] In S234, based on the health level of each key limb component, local feature constraint coefficients are configured for the corresponding associated movement features. For example, when the component is at the weak level, the corresponding joint angle, local posture control and other movement features are matched with a constraint coefficient greater than 1, relaxing the restrictions to form a protective assessment; when the component is at the normal level, the constraint coefficient is 1; when the component is at the excellent level, the constraint coefficient is less than 1, tightening the judgment criteria for the corresponding movement features. This coefficient only applies to the specific movement features associated with the limb and does not affect unrelated movement features.

[0037] In S235, training fatigue is determined based on real-time status data, including cumulative training time, percentage of movement deformation, amplitude of movement rhythm decay, and degree of body swaying. If the training fatigue is higher, it indicates more obvious movement degradation, and the correction coefficient is larger, so the overall evaluation threshold should be temporarily relaxed. If the training fatigue is lower, it indicates that the training status is stable and the movement performance is stable, so the correction coefficient is close to 1, preserving the original benchmark judgment standard to the greatest extent.

[0038] In S26, the overall basic adjustment coefficient, local feature constraint coefficient, and temporary elastic correction coefficient are multiplied together to obtain the comprehensive control coefficient. The comprehensive control coefficient is then multiplied by the upper and lower limits of the benchmark threshold range to scale the original benchmark threshold range. Enlarging the range means widening the tolerance, while narrowing the range means tightening the standard. Finally, a personalized threshold range is output that adapts to the user's physical fitness, the health status of body parts, and the current real-time training status.

[0039] This solution integrates multi-dimensional adjustment coefficients based on fitness level, limb health status, and real-time training status to dynamically quantify and correct the baseline threshold range. It can adaptively relax or tighten the assessment criteria according to different users' exercise foundation, physical differences, and real-time training status. This effectively solves the problem that a uniform fixed threshold cannot be adapted to the differences in users' fitness level, limb flexibility, and individual physical conditions. It avoids the defects of inaccurate judgment and poor adaptability caused by a one-size-fits-all assessment, and significantly improves the rationality, personalization, and scoring accuracy of multi-dimensional movement assessment.

[0040] In S24, refer to Figure 4 The flowchart illustrating the method for obtaining motion feature values ​​shows that "extracting user motion feature values ​​from the dynamic skeleton map of the current action stage in real time according to multiple evaluation dimensions" includes the following steps: S241: Real-time analysis of the dynamic skeleton diagram corresponding to the current action stage to determine the target joints and core stability key points required to complete the current target action; S242: Acquire spatial angle data of the joints used by the target, temporal data of continuous frame motion changes, and real-time coordinate offset data of the core stabilization key points within a preset time period; S243: Use the spatial angle data of the joints of each target as the motion feature value of the corresponding limb posture feature under the posture evaluation dimension; S244: Based on the continuous frame motion change time-series data of each target's joints, obtain the motion feature values ​​of the corresponding limb movement rhythm features under the time-series evaluation dimension; S245: Based on the coordinate offset data of the core stability key points, obtain the motion feature values ​​of the corresponding limb sway control features under the stability assessment dimension.

[0041] In S241, a lightweight real-time posture assessment technology is employed to analyze the dynamic skeletal map of the current action phase (such as a series of squats) frame by frame. This accurately determines the core execution parts of the current target action (a squatting action) and filters out the target joints necessary to complete the action. Taking the squatting action as an example, the target joints include the knee joint, hip joint, ankle joint, and lumbar and back joints. Simultaneously, the core stability key points required to maintain the stability of this action are identified, specifically the pelvic center point, the trunk midline key point, and the shoulder symmetry key point.

[0042] In S242, for the identified target joints, the 3D coordinates of key points of the limbs connected to each joint in each frame of the skeletal image are extracted in real time. Using a spatial vector angle calculation method, the real-time spatial angle data of each joint used by the target is calculated and recorded (e.g., the knee flexion angle and hip extension angle in a squatting motion). The coordinate change trajectory of each joint used by the target is simultaneously acquired in consecutive frames, generating continuous frame motion change time-series data, including the rate of change of joint angles and the time-series curve of motion displacement. For core stabilization key points, a preset duration adapted to the current motion stage is set (e.g., 0.1 seconds per frame for a squatting motion, preset duration 0.5 seconds). The 3D coordinate changes of each stabilization key point within this duration are recorded in real time, and the real-time coordinate offset data of each key point is calculated and obtained, reflecting the positional fluctuations of the body during the motion.

[0043] In S243, the spatial angle data of the joints is used for each target without any standard comparison or deviation calculation. It is directly used as the original motion feature value of the limb posture feature corresponding to the joint in the posture evaluation dimension. Taking the squatting action as an example, the real-time bending angle data of the knee joint is used as the motion feature value of the knee joint posture feature, and the real-time extension angle data of the hip joint is used as the motion feature value of the hip joint posture feature, so as to accurately quantify the static posture state of each limb part in the current target action.

[0044] In S244, based on the continuous frame motion change time series data of the target's joints, core indicators such as action execution rate and action phase switching duration are extracted. Taking the squatting action as an example, the descent speed and time series curve of the squatting action are calculated by the rate of change of the knee joint bending angle in continuous frames, and the action rhythm of the standing phase is calculated by the time series of the hip joint extension angle change. These quantitative indicators are used as action feature values ​​of the corresponding limb movement rhythm characteristics under the time series evaluation dimension, accurately reflecting the execution rhythm and speed of the current target action.

[0045] In S245, the coordinate offset data of the core stability key points are statistically calculated to obtain the maximum value, offset variance and average offset of each key point within a preset time. Taking the squatting action as an example, the coordinate offset data of the pelvic center point is used to quantify the left and right swaying and forward and backward tilting of the body during the squatting process. These statistical data are used as the action feature values ​​of the corresponding limb swaying control characteristics under the stability assessment dimension, which intuitively reflects the user's core control ability of the current target action.

[0046] In S25, the result of matching the action feature value of each evaluation dimension in the current action phase with the corresponding personalized threshold range includes two types: the action feature value is within the corresponding personalized threshold range and the action feature value is not within the corresponding personalized threshold range.

[0047] In S26, refer to Figure 5 The flowchart illustrating the scoring method, "Based on the matching results, obtaining the scoring value for each action feature under each evaluation dimension," includes the following steps: S261: When the action feature value is within the corresponding personalized threshold range, the corresponding score value is set to the preset maximum value; S262: When the action feature value is not within the corresponding personalized threshold range, find the boundary value that is closest to the action feature value from the personalized threshold range; S263: Obtain the corresponding score based on the difference between the nearest boundary value and the action feature value.

[0048] The preset maximum value is 100 points. When the action feature value falls completely within the upper and lower limits of the personalized threshold for the user, it is determined that the action performance of the limb feature conforms to the personalized action specification, and the preset full score is directly assigned to the feature item.

[0049] Boundary values ​​refer to the upper and lower limits of the personalized threshold range. When the action feature value is higher than the upper threshold, the upper threshold is selected as the closest boundary value. When the action feature value is lower than the lower threshold, the lower threshold is selected as the closest boundary value.

[0050] The absolute deviation of the action feature value from the nearest boundary value is calculated. The relative deviation percentage is then calculated using the nearest boundary value as the denominator. This percentage is then adjusted using a weighting coefficient to adjust the overall impact of the deviation. A linear deduction is performed using the full percentage score as the base, and a lower limit on the score is constrained to avoid negative scores. The final score is then output. The formula for calculating the score is as follows: In the formula, Indicates the first The first evaluation dimension The score value of each action feature; Indicates the highest value; Indicates the first The first evaluation dimension Action feature values ​​of each action feature; Indicates the relationship with the first The first evaluation dimension The boundary value that the action feature is closest to; This represents the weighting coefficient.

[0051] In S3, when the score value corresponding to any action feature under any evaluation dimension is lower than the preset maximum value, it is determined that there is an action deviation problem for the action feature. Combining the action stage to which the current target action belongs, the direction of the action feature deviation, and the deviation range between the action feature value and the nearest boundary value, it matches the pre-stored standardized error description and error type label, such as joint angle too small under the posture dimension, action rhythm too fast under the temporal dimension, and limb swaying amplitude too large under the stability dimension, to accurately locate and output a concrete description of the action error type and deviation degree.

[0052] In S4, refer to Figure 6 The flowchart illustrating the guidance strategy acquisition method shows that "based on preset user profiles and error types, determining and providing appropriate guidance strategies to users in real time" includes the following steps: S41: Extract user profile parameters from user profiles; S42: Construct query conditions based on user profile parameters, error types, and corresponding rating values; S43: Based on the query conditions, match the appropriate guidance strategy from the guidance strategy rule base.

[0053] In S41, user profile parameters that can be used for strategic decision-making are extracted from the user profile, including fitness level, physical condition information, fitness goals, age range, learning preferences, and historical error-prone points, etc.

[0054] In S42, the specific error type output by the action evaluation stage and the corresponding score value are used as the basis for action problems. The user profile parameters extracted above are combined and encapsulated to form structured query conditions. These query conditions integrate action defect features and user individual difference attributes to achieve the combination and binding of multi-dimensional constraints, which can improve retrieval accuracy.

[0055] In S43, a pre-built guidance strategy rule base is constructed. This rule base stores a large number of decision rules centered on "condition-action," with each rule bound to a specific trigger condition and a corresponding guidance action. The query conditions are traversed and compared logically with each decision rule in the guidance strategy rule base to select the optimal guidance strategy that fits the current error type, score performance, and user profile characteristics. The prompt text, feedback tone, and display format are determined, and the accompanying control logic such as the subsequent demonstration pace and training difficulty can be adjusted as needed.

[0056] For example, if the error type is identified as insufficient range of motion or a posture dimension score below the maximum, and the user profile parameters indicate a beginner level of physical fitness, this error type, low score, and beginner profile parameters together constitute the query conditions. A beginner-specific encouraging guidance rule is matched against the guidance strategy rule base. The difference between the user's action feature value and the nearest boundary value—that is, the score—is directly used as the range of motion adjustment, thereby generating encouraging guidance messages with precise numerical values, such as "Great movement! Try squatting down 5 centimeters further and feel the thigh muscles engage." The guidance content can take the form of text, voice, highlighted annotations, and comparative animations, facilitating the provision of the most intuitive and effective correction solutions.

[0057] In S5, the actual effect of the guidance strategy is comprehensively and quantitatively evaluated by combining the fluctuations of real-time scores of each dimension during a single training session, the frequency of occurrence of various types of action errors, the number of repeated errors, and the degree of action improvement after guidance prompts. By comparing the degree of reduction in user action feature deviations, the magnitude of score recovery, and whether similar errors have decreased before and after guidance is pushed, the user's acceptance of action correction guidance and the effectiveness of action correction are objectively determined, forming a guidance effect evaluation result that can be used for profile iteration.

[0058] Reference Figure 7 The flowchart illustrating the user profile update method shows that "updating user profiles based on guidance effectiveness" includes the following steps: S51: Based on the guidance effect, update the user's fitness level and physical condition information; S52: Update user profiles based on new fitness level and physical condition information.

[0059] In S51, based on long-term accumulated multi-round guidance effect evaluation data, overall average movement score, movement stability performance and high-frequency error distribution patterns, and matching preset advancement rules and ability correction rules, if the user's movement standard and stability continue to meet the standards and errors are significantly reduced after targeted guidance, the physical fitness level will be gradually increased; if similar limb-related errors recur after guidance and the score of specific joint movements remains low, the corresponding limb's health level and physical condition label will be marked and fine-tuned simultaneously to objectively reflect the user's true athletic ability and limb adaptation status.

[0060] In S52, the adjusted fitness level and optimized physical condition information are synchronously written into and overwrite the corresponding parameter modules of the user profile, completing the iterative update of the profile's core static parameters. The updated user profile can be directly called upon in the next round of adaptive adjustment of action thresholds and matching of guidance strategy rules. This continuous cycle of effect evaluation and data iteration ensures that the user profile dynamically adapts to the long-term changes in the user's athletic ability, achieving a closed-loop linkage between training evaluation, personalized guidance, and dynamic optimization of the profile.

[0061] Reference Figure 8 This disclosure provides a personalized practice guidance system, including: The real-time posture assessment module 101 is used to collect the user's follow-up video in real time, perform real-time posture assessment on the follow-up video, and obtain a dynamic skeleton map. The user action evaluation module 102 is used to evaluate the user's actions in multiple dimensions based on the skeleton diagram and standard action video library, and obtain the score value of each evaluation dimension. Error type identification module 103 is used to identify the user's error type based on the score value of each evaluation dimension; The guidance strategy determination module 104 is used to determine and provide appropriate guidance strategies to users in real time based on preset user profiles and error types. The user profile update module 105 is used to evaluate the guidance effect based on the rating value and error type, and update the user profile based on the guidance effect.

[0062] The various variations and specific examples of the personalized follow-up guidance method provided above are also applicable to the personalized follow-up guidance system provided in this disclosure. Through the foregoing detailed description of the personalized follow-up guidance method, those skilled in the art can clearly understand the implementation method of the personalized follow-up guidance system. For the sake of brevity, they will not be described in detail here.

[0063] A computer device according to embodiments of the present disclosure includes a memory and a processor. The memory is used to store non-transitory computer-readable instructions. Specifically, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, etc.

[0064] The processor may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the computer device to perform desired functions. In one embodiment of this disclosure, the processor is used to execute computer-readable instructions stored in the memory, causing the computer device to perform all or part of the steps of the personalized follow-up guidance methods of the foregoing embodiments of this disclosure.

[0065] Those skilled in the art will understand that, in order to solve the technical problem of how to achieve a good user experience, this embodiment may also include well-known structures such as communication buses and interfaces, and these well-known structures should also be included within the protection scope of this disclosure.

[0066] like Figure 9 This is a schematic diagram of a computer device provided for an embodiment of the present disclosure. It illustrates a structural schematic diagram suitable for implementing the computer device in the embodiments of the present disclosure. Figure 9 The computer device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0067] like Figure 9 As shown, a computer device may include a processor (such as a central processing unit, graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) or programs loaded from storage devices into random access memory (RAM). The RAM also stores various programs and data required for the operation of the computer device. The processor, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0068] Typically, the following devices can be connected to the I / O interface: input devices, such as sensors or visual information acquisition devices; output devices, such as displays; storage devices, such as magnetic tapes or hard drives; and communication devices. Communication devices allow the computer device to communicate wirelessly or wiredly with other devices (such as edge computing devices) to exchange data. Although Figure 9 A computer apparatus with various devices is shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or included alternatively.

[0069] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a processor, all or part of the steps of the personalized follow-up guidance method of embodiments of this disclosure are performed.

[0070] For a detailed description of this embodiment, please refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0071] A computer-readable storage medium according to embodiments of the present disclosure stores non-transitory computer-readable instructions. When the non-transitory computer-readable instructions are executed by a processor, all or part of the steps of the personalized follow-up guidance methods of the foregoing embodiments of the present disclosure are performed.

[0072] The aforementioned computer-readable storage media include, but are not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or portable hard drive), media with built-in rewritable non-volatile memory (e.g., memory card), and media with built-in ROM (e.g., ROM cartridge).

[0073] For a detailed description of this embodiment, please refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0074] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this disclosure are merely examples and not limitations, and should not be considered as essential features of each embodiment of this disclosure. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the scope of this disclosure to the necessity of employing the aforementioned specific details for implementation.

[0075] In this disclosure, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The block diagrams of devices, apparatuses, devices, and systems involved in this disclosure are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as "comprising," "including," "having," etc., are open-ended terms meaning "including but not limited to," and are used interchangeably with them. The terms "or" and "and" as used herein refer to the terms "and / or," and are used interchangeably with them unless the context clearly indicates otherwise. The term "such as" as used herein refers to the phrase "such as but not limited to," and is used interchangeably with it.

[0076] Additionally, as used herein, the "or" used in a list of items beginning with "at least one" indicates a separate list, such that a list of, for example, "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not imply that the described example is preferred or better than other examples.

[0077] It should also be noted that in the systems and methods of this disclosure, the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered as equivalent solutions to this disclosure.

[0078] Various changes, substitutions, and modifications can be made to the technology described herein without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of this disclosure is not limited to the specific aspects of the processes, machines, manufactures, events, means, methods, and actions described above. Currently existing or later-developed processes, machines, manufactures, events, means, methods, or actions that perform substantially the same function or achieve substantially the same result as the corresponding aspects described herein can be utilized. Therefore, the appended claims include such processes, machines, manufactures, events, means, methods, or actions within their scope.

[0079] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0080] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A personalized follow-up coaching method, characterized by, include: The system collects users' practice videos in real time, performs real-time posture evaluation on the practice videos, and obtains dynamic skeletal diagrams. Based on the skeletal diagram and standard motion video library, the user's actions are evaluated in multiple dimensions, and a score value is obtained for each evaluation dimension. Based on the score value of each evaluation dimension, identify the user's error type; Based on the preset user profile and the error type, determine and provide appropriate guidance strategies to the user in real time; The effectiveness of the guidance is evaluated based on the rating and the error type, and the user profile is updated based on the effectiveness of the guidance.

2. The personalized follow-up guidance method according to claim 1, characterized in that, Based on the skeletal diagram and standard motion video library, the user's movements are evaluated in multiple dimensions to obtain a score for each evaluation dimension, including: Extract target standard action videos from the standard action video library, segment the target standard action videos according to action stages, and obtain multiple action stage sub-videos; Obtain the baseline threshold range for each action feature in each action stage sub-video across multiple evaluation dimensions; Based on the user profile, the baseline threshold range is dynamically adjusted to obtain a personalized threshold range that suits the user. Based on multiple evaluation dimensions, the user's action feature values ​​are extracted in real time from the dynamic skeleton map of the current action stage; Match the action feature values ​​of each evaluation dimension in the current action phase with the corresponding personalized threshold range; Based on the matching results, obtain the score value for each action feature under each evaluation dimension.

3. The personalized follow-up training guidance method according to claim 2, characterized in that, The step of dynamically adjusting the baseline threshold range based on the user profile to obtain a personalized threshold range suitable for the user includes: Based on the user profile, the user's physical fitness level and physical condition information are obtained; based on the training video, the user's real-time status data is obtained. Based on the aforementioned physical condition information, the health level of each key limb component of the user is obtained; An overall basic adjustment coefficient is set based on the aforementioned physical fitness level. Based on the health level of each key limb component, local feature constraint coefficients are set for each movement feature under multiple evaluation dimensions; A temporary elastic correction coefficient is set based on the real-time status data; Based on the overall basic adjustment coefficient, the local feature constraint coefficient, and the temporary elastic correction coefficient, the benchmark threshold range is dynamically adjusted to obtain a personalized threshold range suitable for the user.

4. The personalized follow-up guidance method according to claim 2, characterized in that, The evaluation dimensions include at least one of posture, temporality, and stability; the extraction of user action feature values ​​from the dynamic skeleton map of the current action stage in real time according to multiple evaluation dimensions includes: Real-time analysis of the dynamic skeleton map corresponding to the current action stage to determine the target joints and core stability key points required to complete the current target action; Acquire spatial angle data of the joints used by the target, temporal data of continuous frame motion changes, and real-time coordinate offset data of the core stabilization key points within a preset time period; The spatial angle data of the joints of each target are used as the motion feature values ​​of the corresponding limb posture features under the posture evaluation dimension; Based on the continuous frame motion change time-series data of the joints used by each target, the motion feature values ​​of the corresponding limb motion rhythm features under the time-series evaluation dimension are obtained. Based on the coordinate offset data of the core stability key points, the motion feature values ​​of the corresponding limb sway control features under the stability assessment dimension are obtained.

5. The personalized follow-up guidance method according to claim 2, characterized in that, The process of obtaining a score for each action feature under each evaluation dimension based on the matching results includes: When the action feature value is within the corresponding personalized threshold range, the corresponding score value is set to the preset maximum value; When the action feature value is not within the corresponding personalized threshold range, find the boundary value that is closest to the action feature value from the personalized threshold range; The corresponding score is obtained based on the difference between the nearest boundary value and the action feature value.

6. The personalized follow-up guidance method according to claim 1, characterized in that, The process of determining and providing appropriate guidance strategies to users in real time based on preset user profiles and error types includes: Extract user profile parameters from the user profile; Based on the user profile parameters, the error type, and the corresponding score, construct the query conditions; Based on the query conditions, a suitable guidance strategy is matched from the guidance strategy rule base.

7. The personalized follow-up training guidance method according to claim 1, characterized in that, The step of updating the user profile based on the guidance effect includes: Based on the guidance effect, update the user's physical fitness level and physical condition information; The user profile is updated based on the new fitness level and physical condition information.

8. A computer device, characterized in that, The computer device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the personalized follow-up guidance method according to any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to perform the personalized follow-up guidance method as described in any one of claims 1-7.

10. A computer program product comprising computer instructions, characterized in that, When executed by a processor, the computer instructions implement the steps of the personalized follow-up guidance method according to any one of claims 1-7.

Images