Construction method of fitness training evaluation decision model based on VR virtual digital human

Through high-precision motion capture and biomechanical analysis, combined with computer vision technology, a fitness training evaluation decision model based on a decision tree model is constructed, which solves the problem of individual differences assessment and realizes accurate fitness training evaluation and guidance.

CN119027558AActive Publication Date: 2024-11-26BEIJINGS POWER TO EDUCATION TECH CO LTD

Patent Information

Application Number
CN202411098040.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-26
Estimated Expiration
2044-08-12

AI Technical Summary

Technical Problem

The existing fitness training evaluation decision-making model based on VR virtual digital people is difficult to effectively evaluate the differences in individual body shape, flexibility and cardiopulmonary function. The difficulty of traditional exercises is not suitable for different individuals, and there is a lack of precise evaluation and guidance.

Method used

Expert action data is captured through a high-precision inertial camera group network, and biomechanical parameters are analyzed using rigid body dynamics model and musculoskeletal simulation, combined with computer vision technology to identify students' actions, generate training evaluation reports, and build an evaluation decision model based on the decision tree model.

Benefits of technology

Accurate evaluation and guidance of individual movements is achieved, targeted training suggestions are provided, training effect and efficiency are improved, and evaluation errors are reduced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119027558A_ABST
    Figure CN119027558A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of computers, in particular to a building method of a fitness training evaluation decision model based on a VR virtual digital human. The method comprises the following steps: performing motion capture on exercise actions of an expert through a high-precision inertial camera group network to obtain expert action data which comprises joint motion angle data and muscle activity potential data; performing joint torque, tendon transmission ratio and inertia force analysis on the expert action data based on a rigid body dynamic model and musculoskeletal simulation to obtain biomechanical parameters; and importing the expert action data and the biomechanical parameters into a virtual reality engine, and performing virtual digital human action demonstration generation through skeleton driving and inverse kinematics to obtain virtual action demonstration data. Biomechanical models such as a rigid body dynamic model and a musculoskeletal model are fused, and joint torque, tendon transfer ratio and inertia force biological parameters are quantitatively calculated, so that simulation is more real.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method for constructing a fitness training evaluation decision model based on VR virtual digital human. Background Art

[0002] With the rapid development of virtual reality (VR), augmented reality (AR), artificial intelligence (AI) and motion capture technology, people have gradually begun to apply these technologies to the fields of fitness and sports training. Traditional fitness training evaluation methods often rely on manual guidance and experience accumulation, lacking objective and accurate evaluation methods; this method is not only inefficient, but also prone to evaluation errors, affecting the training effect. In this context, a fitness training evaluation decision model based on VR virtual digital humans came into being, aiming to use advanced technical means to accurately evaluate and guide fitness training, thereby improving the training effect.

[0003] However, there are still some problems in the current construction method of fitness training evaluation decision-making model based on VR virtual digital human: different individuals have significant differences in body shape, flexibility, cardiopulmonary function, etc., and the difficulty of traditional exercises is also different for different individuals; how to evaluate the physical characteristics of individuals and provide targeted exercise suggestions based on individual differences is an important issue that needs to be solved. Traditional exercises usually involve fine and smooth body movements, which require accurate capture and modeling of expert movement data; this places high demands on the accuracy of motion capture equipment and the robustness of motion recognition algorithms. Summary of the invention

[0004] Based on this, it is necessary for the present invention to provide a method for constructing a fitness training evaluation decision model based on VR virtual digital human to solve at least one of the above technical problems.

[0005] To achieve the above purpose, a method for constructing a fitness training evaluation decision model based on VR virtual digital human includes the following steps:

[0006] Step S1: capturing the expert's practice movements through a high-precision inertial camera group network to obtain expert movement data, wherein the expert movement data includes joint movement angle data and muscle activity potential data;

[0007] Step S2: Analyze the joint torque, tendon transfer ratio and inertial force of the expert action data based on the rigid body dynamics model and musculoskeletal simulation to obtain biomechanical parameters;

[0008] Step S3: importing the expert motion data and biomechanical parameters into the virtual reality engine, and performing virtual digital human motion demonstration generation through skeletal drive and inverse kinematics to obtain virtual motion demonstration data;

[0009] Step S4: using computer vision technology to enhance the feature expression of key joints of the pre-acquired student action video data, and performing exercise action recognition on the student action video data according to the virtual action demonstration data, thereby obtaining the student movement pattern data;

[0010] Step S5: Perform human biological parameter analysis on the student's action video data to obtain the student's biological parameters; perform difference measurement analysis on the student's motion pattern data based on the virtual action demonstration data to obtain action difference measurement data; perform action adjustment based on the biomechanical model according to the student's biological parameters and the action difference measurement data, and generate training evaluation report data; construct a training evaluation decision model based on the decision tree model according to the training evaluation report data.

[0011] The present invention can accurately obtain the expert's motion data, including joint motion angles and muscle action potential data, which are the basis for subsequent analysis and simulation. Through biomechanical analysis, parameters such as joint torque, tendon transfer ratio and inertial force can be obtained. These parameters are very important for subsequent action generation and evaluation, and can provide specific biomechanical guidance. The virtual action demonstration data presents the simulation results of the expert's actions through a virtual reality engine. This visualization method can more intuitively show the expert's action skills and provide specific demonstrations and references for students. Through computer vision technology, the feature expression of key joint points can be extracted from the student's action video, and the understanding and analysis of the student's actions can be enhanced. By comparing with the virtual action demonstration data, the identification of the exercise action and the data extraction of the student's movement pattern can be carried out. By analyzing the biological parameters of the trainees' action video data, the biological parameters of the trainees, such as the range of joint motion and the degree of muscle activity, can be obtained; based on the difference measurement analysis of the virtual action demonstration data and the trainees' motion pattern data, the difference between the trainees' actions and the expert's actions can be evaluated; according to the trainees' biological parameters and action difference measurement data, the action adjustment based on the biomechanical model can be carried out, and targeted training suggestions and optimization plans can be provided; finally, based on the training evaluation report data, a training evaluation decision model based on the decision tree model is constructed, which can automatically evaluate and make decisions on the trainees' training performance and provide guidance and feedback for further training. In summary, the effects of the above steps include: providing expert action data and biomechanical parameters to provide a basis for subsequent analysis and simulation; through virtual action demonstration data, the expert action skills are intuitively displayed to provide reference and demonstration for trainees; using computer vision technology and virtual action demonstration data, the trainees' actions are analyzed and identified, and the trainees' motion pattern data are extracted; analyzing the trainees' human biological parameters and action difference measurement data, the action adjustment based on the biomechanical model is carried out, and the training plan is optimized; and a training evaluation decision model based on the decision tree model is constructed to automatically evaluate the trainees' performance and provide guidance and feedback. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0013] Figure 1 It is a schematic diagram of the steps of the method for constructing a fitness training evaluation decision model based on VR virtual digital human in the present invention;

[0014] Figure 2 for Figure 1 Detailed step flow diagram of step S1;

[0015] Figure 3 for Figure 1 Detailed step flow chart of step S2 in FIG. DETAILED DESCRIPTION

[0016] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0017] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0018] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0019] To achieve this, please refer to Figures 1 to 3 The present invention provides a method for constructing a fitness training evaluation decision model based on VR virtual digital human, the method comprising the following steps:

[0020] Step S1: capturing the expert's practice movements through a high-precision inertial camera group network to obtain expert movement data, wherein the expert movement data includes joint movement angle data and muscle activity potential data;

[0021] Step S2: Analyze the joint torque, tendon transfer ratio and inertial force of the expert action data based on the rigid body dynamics model and musculoskeletal simulation to obtain biomechanical parameters;

[0022] Step S3: importing the expert motion data and biomechanical parameters into the virtual reality engine, and performing virtual digital human motion demonstration generation through skeletal drive and inverse kinematics to obtain virtual motion demonstration data;

[0023] Step S4: using computer vision technology to enhance the feature expression of key joints of the pre-acquired student action video data, and performing exercise action recognition on the student action video data according to the virtual action demonstration data, thereby obtaining the student movement pattern data;

[0024] Step S5: Perform human biological parameter analysis on the student's action video data to obtain the student's biological parameters; perform difference measurement analysis on the student's motion pattern data based on the virtual action demonstration data to obtain action difference measurement data; perform action adjustment based on the biomechanical model according to the student's biological parameters and the action difference measurement data, and generate training evaluation report data; construct a training evaluation decision model based on the decision tree model according to the training evaluation report data.

[0025] In the embodiment of the present invention, reference Figure 1 The above is a schematic diagram of the steps of a method for constructing a fitness training evaluation decision model based on a VR virtual digital human in the present invention. In this example, the method for constructing a fitness training evaluation decision model based on a VR virtual digital human includes the following steps:

[0026] Step S1: capturing the expert's practice movements through a high-precision inertial camera group network to obtain expert movement data, wherein the expert movement data includes joint movement angle data and muscle activity potential data;

[0027] The embodiment of the present invention uses a high-precision inertial camera group network for motion capture. This network usually includes multiple inertial sensors, such as accelerometers and gyroscopes, which can accurately measure the position, posture and movement of an object; at the same time, a muscle activity potential sensor is used to capture the electrical signals of the muscle to obtain muscle activity data.

[0028] Step S2: Analyze the joint torque, tendon transfer ratio and inertial force of the expert action data based on the rigid body dynamics model and musculoskeletal simulation to obtain biomechanical parameters;

[0029] The embodiments of the present invention use rigid body dynamics models and musculoskeletal simulation technology to analyze expert motion data. These models and simulation methods can calculate biomechanical parameters such as joint torque, tendon transfer ratio and inertial force based on factors such as joint angle, muscle activity and object mass.

[0030] Step S3: importing the expert motion data and biomechanical parameters into the virtual reality engine, and performing virtual digital human motion demonstration generation through skeletal drive and inverse kinematics to obtain virtual motion demonstration data;

[0031] The embodiments of the present invention use a virtual reality engine, such as Unity or Unreal Engine, to import expert motion data and biomechanical parameters, and use skeletal drive and inverse kinematics algorithms to generate motion demonstrations of virtual digital humans; these engines provide powerful graphics rendering and animation generation capabilities, and can present realistic virtual motion demonstrations.

[0032] Step S4: using computer vision technology to enhance the feature expression of key joints of the pre-acquired student action video data, and performing exercise action recognition on the student action video data according to the virtual action demonstration data, thereby obtaining the student movement pattern data;

[0033] The embodiment of the present invention uses computer vision technology to analyze the student's action video data, and can use deep learning models, such as convolutional neural networks (CNN), to extract features and enhance the expression of key joints in the video frames; at the same time, with the help of virtual action demonstration data, the student's action video can be used to identify the exercises based on the comparison and matching methods, thereby obtaining the student's movement pattern data.

[0034] Step S5: Perform human biological parameter analysis on the student's action video data to obtain the student's biological parameters; perform difference measurement analysis on the student's motion pattern data based on the virtual action demonstration data to obtain action difference measurement data; perform action adjustment based on the biomechanical model according to the student's biological parameters and the action difference measurement data, and generate training evaluation report data; construct a training evaluation decision model based on the decision tree model according to the training evaluation report data.

[0035] The embodiment of the present invention uses human biological parameter analysis technology to process the student's action video data, and can extract the student's biological parameters, such as joint angles, muscle activity and other information; at the same time, using virtual action demonstration data and student movement pattern data, a difference measurement analysis can be performed to compare the differences between the student's actions and the expert's actions. Based on the biomechanical model, combined with the student's biological parameters and action difference measurement data, action adjustments can be made, that is, corresponding guidance and training adjustments are made according to the student's physical characteristics and action differences; finally, training evaluation report data is generated, including the student's training results and suggestions. The training evaluation report data can be analyzed and decided using a decision tree model to construct a training evaluation decision model.

[0036] The present invention can accurately obtain the expert's motion data, including joint motion angles and muscle action potential data, which are the basis for subsequent analysis and simulation. Through biomechanical analysis, parameters such as joint torque, tendon transfer ratio and inertial force can be obtained. These parameters are very important for subsequent action generation and evaluation, and can provide specific biomechanical guidance. The virtual action demonstration data presents the simulation results of the expert's actions through a virtual reality engine. This visualization method can more intuitively show the expert's action skills and provide specific demonstrations and references for students. Through computer vision technology, the feature expression of key joint points can be extracted from the student's action video, and the understanding and analysis of the student's actions can be enhanced. By comparing with the virtual action demonstration data, the identification of the exercise action and the data extraction of the student's movement pattern can be carried out. By analyzing the biological parameters of the trainees' action video data, the biological parameters of the trainees, such as the range of joint motion and the degree of muscle activity, can be obtained; based on the difference measurement analysis of the virtual action demonstration data and the trainees' motion pattern data, the difference between the trainees' actions and the expert's actions can be evaluated; according to the trainees' biological parameters and action difference measurement data, the action adjustment based on the biomechanical model can be carried out, and targeted training suggestions and optimization plans can be provided; finally, based on the training evaluation report data, a training evaluation decision model based on the decision tree model is constructed, which can automatically evaluate and make decisions on the trainees' training performance and provide guidance and feedback for further training. In summary, the effects of the above steps include: providing expert action data and biomechanical parameters to provide a basis for subsequent analysis and simulation; through virtual action demonstration data, the expert action skills are intuitively displayed to provide reference and demonstration for trainees; using computer vision technology and virtual action demonstration data, the trainees' actions are analyzed and identified, and the trainees' motion pattern data are extracted; analyzing the trainees' human biological parameters and action difference measurement data, the action adjustment based on the biomechanical model is carried out, and the training plan is optimized; and a training evaluation decision model based on the decision tree model is constructed to automatically evaluate the trainees' performance and provide guidance and feedback.

[0037] Preferably, step S1 comprises the following steps:

[0038] Step S11: using texture-free blue-green as the background, arranging multiple high-precision inertial capture cameras to cover the full view, and obtaining a high-precision inertial camera group network;

[0039] Step S12: Paste reflective markers on key parts of the expert's body and wear a surface electromyography sensor. The expert performs a complete demonstration of the exercise action process in the inertial camera group network to collect the original electromyography data and original 3D position data of the markers;

[0040] Step S13: eliminating noise and missing information from the original electromyographic data and the original 3D position data, and synchronizing the timestamps of the position and electromyographic data, thereby obtaining muscle activity data and 3D marker position data;

[0041] Step S14: constructing a skeleton model according to the muscle activity data and the 3D marker point position data by using biomechanical modeling software, thereby obtaining expert skeleton model data;

[0042] Step S15: Bind the 3D marker point position data to the expert skeleton model data to calculate the joint motion angle, thereby obtaining the joint motion angle data; estimate the muscle activity level and action intensity according to the muscle activity data, thereby obtaining the muscle action potential data;

[0043] Step S16: Store the joint motion angle data and muscle action potential data as expert motion data.

[0044] As an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 Detailed step flow diagram of step S1 in the embodiment of the present invention, step S1 includes the following steps:

[0045] Step S11: using texture-free blue-green as the background, arranging multiple high-precision inertial capture cameras to cover the full view, and obtaining a high-precision inertial camera group network;

[0046] The embodiments of the present invention use multiple high-precision inertial capture cameras, which usually have high resolution and fast sampling rate, and can accurately capture the position, posture and movement of objects. These cameras are arranged to achieve full-view coverage and ensure that the background is texture-free blue-green to facilitate background separation in subsequent processing.

[0047] Step S12: Paste reflective markers on key parts of the expert's body and wear a surface electromyography sensor. The expert performs a complete demonstration of the exercise action process in the inertial camera group network to collect the original electromyography data and original 3D position data of the markers;

[0048] Embodiment of the present invention: Reflective marking points are pasted on key parts of the expert's body. These marking points are usually highly reflective so that the camera can accurately capture their positions; at the same time, the expert also wears a surface electromyography sensor to record the electrical signals of muscle activity. The expert performs a complete demonstration in an inertial camera group network. The camera will capture the position and posture of the marking points and record the electrical signals of the electromyography sensor.

[0049] Step S13: eliminating noise and missing information from the original electromyographic data and the original 3D position data, and synchronizing the timestamps of the position and electromyographic data, thereby obtaining muscle activity data and 3D marker position data;

[0050] The embodiment of the present invention uses signal processing technology to perform noise elimination and filtering operations on the original electromyographic data to remove interference and extract effective muscle activity signals; processes the original 3D position data to remove possible noise and omissions, and aligns the electromyographic data and position data through timestamp synchronization to ensure that they are consistent in time.

[0051] Step S14: constructing a skeleton model according to the muscle activity data and the 3D marker point position data by using biomechanical modeling software, thereby obtaining expert skeleton model data;

[0052] The embodiments of the present invention use biomechanical modeling software, such as AnyBody or OpenSim, to build an expert skeleton model based on muscle activity data and 3D marker point position data. These software provide skeleton modeling and simulation functions, and can calculate the posture and dynamic characteristics of the skeleton based on muscle activity and joint position.

[0053] Step S15: Bind the 3D marker point position data to the expert skeleton model data to calculate the joint motion angle, thereby obtaining the joint motion angle data; estimate the muscle activity level and action intensity according to the muscle activity data, thereby obtaining the muscle action potential data;

[0054] The embodiment of the present invention associates and binds 3D marker point position data with expert skeleton model data to realize the movement of the skeleton model. By calculating the angle changes of each joint in the skeleton model, the joint movement angle data can be obtained; according to the muscle activity data, the muscle activity level and action intensity can be estimated to obtain muscle activity potential data.

[0055] Step S16: Store the joint motion angle data and muscle action potential data as expert motion data.

[0056] The embodiment of the present invention saves the calculated joint motion angle data and muscle activity potential data as expert action data; these data can be stored in an appropriate format, such as a CSV (comma separated value) file or a database, and the stored data can be used for subsequent analysis, research or application.

[0057] The present invention adopts a network composed of a texture-free blue-green background and multiple high-precision inertial cameras, which can provide full-view coverage and reduce background interference, so that high-quality and high-accuracy expert action data can be obtained. By pasting reflective markers on key parts of the expert's body and wearing surface electromyography sensors, the detailed information of the expert's actions can be recorded; the collection of raw electromyographic data and raw 3D position data can provide more comprehensive and accurate expert action data. By eliminating noise and processing missing data for the raw electromyographic data and the raw 3D position data, the quality and accuracy of the data can be improved; the synchronization of timestamps can ensure the consistency and correspondence between muscle activity data and 3D marker position data. Through biomechanical modeling software, an expert's skeletal model can be constructed based on muscle activity data and 3D marker position data; this provides an accurate skeletal structure for subsequent joint motion angle calculation and muscle action potential data estimation. By binding the 3D marker position data with the expert skeletal model data, the joint motion angle data can be calculated; this provides an accurate measure of joint motion for analyzing the expert's action skills; at the same time, based on the muscle activity data, the activity level and action intensity of the muscle can be estimated to obtain muscle action potential data. Storing joint motion angle data and muscle activity potential data as expert action data provides detailed information and features of expert actions, which will be used in subsequent analysis, research and applications, such as imitation, evaluation and optimization of expert actions. In general, the effects of the above steps include: providing high-quality and high-accuracy expert action data; capturing detailed information of expert actions, including muscle activity data and joint motion angle data; removing noise and missing data to improve data quality and accuracy; establishing an expert's skeletal model to provide accurate anatomical structure for subsequent analysis and research; estimating muscle activity level and intensity to provide a deeper understanding of muscle activity; storing expert action data to provide a basis for subsequent analysis, research and application. These effects can provide more accurate, detailed and comprehensive data support for the research, evaluation, optimization and application of expert actions, and promote the development and application of related fields.

[0058] Preferably, step S2 comprises the following steps:

[0059] Step S21: constructing a rigid body chain model and a muscle model according to the expert action data; mapping the muscle action potential data into the activity parameters of the muscle model, thereby obtaining a parameterized muscle model;

[0060] Step S22: using the joint motion angle data as motion input to drive the rigid body chain model to move, and solving the force and torque of each joint based on Newton's second law to obtain the torque value of each joint at each time step, thereby obtaining the joint torque data;

[0061] Step S23: using the joint torque data as input, inversely solving the muscle contraction force of the parameterized muscle model, thereby obtaining the muscle contraction force data;

[0062] Step S24: Calculate the moment acting on the bone based on the position of the muscle attachment point according to the muscle contraction force data, so as to obtain the bone moment data; calculate the tendon transfer ratio data according to the bone moment data and the muscle contraction force data, wherein the tendon transfer ratio = bone moment / muscle contraction force;

[0063] Step S25: decomposing the joint motion angle data into the motion of each rigid body according to the rigid body chain model, and calculating the linear acceleration and angular acceleration based on the rigid body, thereby obtaining the rigid body motion parameter data;

[0064] Step S26: deriving the inertial force and equilibrium moment of each body part based on Newton's law of motion from the rigid body motion parameter data, thereby obtaining inertial force data;

[0065] Step S27: Perform bio-parameter analysis based on the joint torque data, tendon transfer ratio data and inertial force data to obtain biomechanical parameters.

[0066] As an embodiment of the present invention, refer to Figure 3 As shown, Figure 1 Detailed step flow diagram of step S2 in the embodiment of the present invention, step S2 includes the following steps:

[0067] Step S21: constructing a rigid body chain model and a muscle model according to the expert action data; mapping the muscle action potential data into the activity parameters of the muscle model, thereby obtaining a parameterized muscle model;

[0068] The embodiment of the present invention uses biomechanical modeling software, such as AnyBody or OpenSim, to construct a rigid body chain model and a muscle model based on expert motion data. The rigid body chain model represents the structural relationship between various parts of the body (such as bones and joints), while the muscle model represents the position, shape and mechanical properties of the muscle. By mapping the muscle activity potential data into the activity parameters of the muscle model, the contraction and relaxation process of the muscle can be simulated.

[0069] Step S22: using the joint motion angle data as motion input to drive the rigid body chain model to move, and solving the force and torque of each joint based on Newton's second law to obtain the torque value of each joint at each time step, thereby obtaining the joint torque data;

[0070] The embodiment of the present invention uses a rigid chain model and joint motion angle data to calculate the force and torque of each joint through Newton's second law (force equals mass multiplied by acceleration); the joint motion angle data is used as motion input to drive the movement of the rigid chain model, and then by solving Newton's second law equation, the torque value of each joint at each time step is obtained, thereby obtaining the joint torque data.

[0071] Step S23: using the joint torque data as input, inversely solving the muscle contraction force of the parameterized muscle model, thereby obtaining the muscle contraction force data;

[0072] The embodiment of the present invention uses a parameterized muscle model and joint torque data to inversely resolve the joint torque into muscle contraction force through inverse calculation; this process involves the mechanical properties of the muscle and the mechanical relationship of the muscle-skeletal system.

[0073] Step S24: Calculate the moment acting on the bone based on the position of the muscle attachment point according to the muscle contraction force data, so as to obtain the bone moment data; calculate the tendon transfer ratio data according to the bone moment data and the muscle contraction force data, wherein the tendon transfer ratio = bone moment / muscle contraction force;

[0074] The embodiment of the present invention uses the muscle contraction force data and the position information of the muscle attachment points in the muscle model to calculate the torque acting on the bones; based on the bone torque data and the muscle contraction force data, the tendon transfer ratio can be calculated, which ratio represents the transmission efficiency of the muscle torque on the bones.

[0075] Step S25: decomposing the joint motion angle data into the motion of each rigid body according to the rigid body chain model, and calculating the linear acceleration and angular acceleration based on the rigid body, thereby obtaining the rigid body motion parameter data;

[0076] The embodiment of the present invention uses a rigid body chain model and joint motion angle data to decompose the joint motion angle into the motion of each rigid body; based on the motion of the rigid body, the linear acceleration and angular acceleration can be calculated, and these parameters describe the motion state of the rigid body.

[0077] Step S26: deriving the inertial force and equilibrium moment of each body part based on Newton's law of motion from the rigid body motion parameter data, thereby obtaining inertial force data;

[0078] The embodiment of the present invention uses Newton's laws of motion, combined with rigid body motion parameter data and information such as the mass and inertia matrix of the rigid body, to derive the inertial forces and equilibrium moments of various parts of the body; these forces and moments are caused by the motion and inertial characteristics of the rigid body.

[0079] Step S27: Perform bio-parameter analysis based on the joint torque data, tendon transfer ratio data and inertial force data to obtain biomechanical parameters.

[0080] The embodiment of the present invention uses joint torque data, tendon transfer ratio data and inertial force data to perform biomechanical parameter analysis; this analysis process involves torque balance, torque transfer efficiency, muscle force distribution, etc., to obtain the evaluation results of biomechanical parameters.

[0081] The present invention can simulate the skeleton and muscle structure of the human body by constructing a rigid chain model and a muscle model, provide a basis for subsequent analysis and calculation, map the muscle action potential data to the activity parameters of the muscle model, quantify the activity level of the muscle, and convert the electrical signal into the contraction degree of the muscle. By taking the joint motion angle data as input, the rigid chain model can be driven to simulate the movement of the human body, and by applying Newton's second law, the force and torque of each joint can be calculated, and the joint torque data are provided; these data can be used to analyze the load and motion characteristics of the joint. By taking the joint torque data as input, the parameterized muscle model can be inversely solved for muscle contraction force, so that the contraction degree and force size of the muscle in a specific action can be inferred, and the muscle contraction force data can be obtained, which are helpful for studying the mechanical behavior and activity characteristics of the muscle. By calculating the torque acting on the bone according to the muscle contraction force data, the influence and load of the muscle on the bone can be understood; at the same time, by calculating the tendon transfer ratio, the transmission efficiency of the muscle force in the skeletal movement can be evaluated, and these data are helpful for understanding the interaction between the muscle and the bone. Through the rigid body chain model, the joint motion angle data can be decomposed into the motion of each rigid body. Based on the calculation of the linear acceleration and angular acceleration of the rigid body, the detailed parameter data of the rigid body motion can be provided. These data are helpful for analyzing and understanding the motion characteristics and coordination of various parts of the body. By applying Newton's laws of motion, the inertial force and equilibrium torque of each part of the body can be derived. These data help to understand the inertial influence of the body during movement and the torque required to maintain balance. The analysis of inertial force can help researchers better understand the dynamic characteristics of human motion. By analyzing the joint torque data, tendon transfer ratio data and inertial force data, biological parameter analysis can be performed. These parameters include joint torque, muscle strength, and mechanical properties of the musculoskeletal system. These biomechanical parameters are of great significance for studying human motion, evaluating exercise load, and optimizing training programs. In summary, by executing the above steps, we can obtain multiple important parameters about human movement, including muscle activity level, joint torque, muscle contraction force, bone torque, tendon transfer ratio, rigid body motion parameters, inertial force, etc. These parameter data have important application value for understanding and analyzing human biomechanical characteristics, movement mechanisms, and optimizing sports training and rehabilitation programs.

[0082] Preferably, step S27 includes the following steps:

[0083] Step S271: Integrate the joint torque data, tendon transfer ratio data and inertial force data into an initial biomechanical parameter set;

[0084] The embodiment of the present invention uses data processing software (such as MATLAB or Python) to create a new data set according to the data format and storage method, which includes the values ​​of joint torque, tendon transfer ratio and inertial force; the collected data is integrated into the data set to ensure the correct matching of the data corresponding to the joint and the action.

[0085] Step S272: Calculate the muscle contraction work according to the joint torque data to obtain muscle contraction work data; calculate the loss work done by the inertial force according to the inertial force data to obtain inertial loss work data;

[0086] The embodiment of the present invention uses a mathematical calculation method to calculate muscle contraction work. According to the joint torque data, the muscle contraction work can be obtained by integrating the joint torque and the speed of the joint angle, and a numerical integration method (such as the trapezoidal rule or the Simpson's rule) can be used for integral calculation. The inertia loss work is calculated using a mathematical calculation method. According to the inertia force data, the inertia loss work can be obtained by calculating the dot product of the inertia force and the linear velocity of the rigid body.

[0087] Step S273: Calculate the total energy during the action according to the muscle contraction work data and the inertia loss work data to obtain energy consumption data, wherein energy consumption = muscle contraction work + inertia loss work;

[0088] The embodiment of the present invention adds the muscle contraction work data and the inertia loss work data to obtain the overall energy consumption data during the action process. Mathematical calculation software or programming language (such as MATLAB or Python) can be used to perform data addition operations to obtain energy consumption data.

[0089] Step S274: Integrate the joint torque data, tendon transfer ratio data, inertial force data and energy consumption data into biomechanical parameters.

[0090] The embodiment of the present invention creates a new data set for integrating joint torque data, tendon transfer ratio data, inertial force data and energy consumption data; integrates these data into the data set in a corresponding format to ensure the correspondence and consistency of the data; finally, the integrated biomechanical parameter data can be saved in an appropriate file format using data processing software or programming language for subsequent analysis and evaluation.

[0091] The present invention integrates data from different sources by integrating joint torque data, tendon transfer ratio data and inertial force data into an initial biomechanical parameter set, so that joint torque, muscle strength, tendon transfer efficiency and inertial influence in body movement can be comprehensively considered, providing comprehensive biomechanical parameters for subsequent analysis. By calculating the joint torque data, the work done by the muscle during the movement, that is, the muscle contraction work, can be derived. This parameter reflects the energy output generated by the muscle during the movement; at the same time, by calculating the work done by the inertial force, that is, the inertial loss work, the energy loss caused by inertia during the movement of the body can be understood. By adding the muscle contraction work data and the inertial loss work data, the overall energy consumption data during the movement can be obtained; this parameter can be used to evaluate the energy consumed by the human body during a specific movement, so as to better understand information such as exercise intensity, training load and energy balance. By integrating joint torque data, tendon transfer ratio data, inertial force data, and energy consumption data into biomechanical parameters, a comprehensive parameter set can be established, covering information on joint torque, muscle strength, tendon transfer efficiency, inertial influence, and energy consumption; this parameter set can be used to study human motion characteristics, evaluate exercise load and energy consumption, and optimize training and rehabilitation programs. In summary, by executing the above steps, more comprehensive and detailed biomechanical parameters about human motion can be obtained, including muscle contraction work, inertial loss work, and energy consumption. These parameters have important application value for studying human motion mechanisms, evaluating exercise load and energy consumption, and optimizing training and rehabilitation programs; integrating these parameters can provide a deeper understanding and analysis, and provide useful information and guidance for research and practice in the field of sports science.

[0092] Preferably, step S3 comprises the following steps:

[0093] Step S31: importing the rigid body chain model and the parameterized muscle model into a virtual reality engine to obtain a virtual human body model;

[0094] The embodiment of the present invention prepares data of a rigid body chain model and a parameterized muscle model, wherein the rigid body chain model describes the human skeleton structure, and the parameterized muscle model describes the shape and motion characteristics of the muscle. A virtual reality engine (such as Unity, Unreal Engine, etc.) is used as a development tool, and the data of the rigid body chain model and the parameterized muscle model are imported into the virtual reality engine; the proportion and posture of the virtual human body model are adjusted to match the real human body.

[0095] Step S32: performing data mapping on the virtual human body model according to the expert motion data, and motion-driving the human body model, thereby obtaining a motion-driven human body model;

[0096] The embodiment of the present invention collects expert motion data, which describes the skeletal movement of a real human body when performing specific actions; uses data processing software (such as MotionBuilder, Blender, etc.) to perform data mapping, and applies the expert motion data to the skeletal structure of a virtual human model; and drives the virtual human model according to the mapped skeletal animation data to simulate the movement in the expert motion data.

[0097] Step S33: generating a high-fidelity skin mesh model based on bones for the action-driven human body model, thereby obtaining a virtual human action model with a skin mesh;

[0098] The embodiment of the present invention uses 3D modeling software (such as Maya, 3ds Max, etc.) to create a mesh model with high-fidelity skin, the shape and appearance of which are similar to those of a real human body; applies skeletal animation to the mesh model with high-fidelity skin to achieve a skin deformation effect that matches the action-driven human body model; ensures the correct matching between the skin mesh and the skeletal animation to ensure the realism of the virtual human body action model.

[0099] Step S34: preliminarily realizing human body movement according to the virtual human body action model through skeletal animation technology, thereby obtaining a virtual action sequence;

[0100] The embodiment of the present invention uses the skeletal animation technology in the virtual reality engine to drive the skeletal animation data of the human body model according to the action to realize the movement of the virtual human body. According to the needs and scenarios, specific action sequences can be defined, including different types of actions such as standing, walking, and running.

[0101] Step S35: performing detailed motion smoothing based on inverse kinematics constraints according to the virtual motion sequence and biomechanical parameters, thereby obtaining a virtual motion optimization sequence;

[0102] The embodiment of the present invention uses an inverse kinematics algorithm to calculate the appropriate posture of the joints based on the key frames and target positions in the virtual action sequence; applies inverse kinematics constraints to ensure that the joints do not exceed their feasible range during movement and have natural movement smoothness; and performs detailed motion smoothing to better simulate the movement details of the real human body.

[0103] Step S36: Perform muscle movement deformation simulation based on biological parameters on the virtual action optimization sequence, so as to obtain virtual action demonstration data with muscle deformation.

[0104] The embodiment of the present invention uses a biomechanical parameter data set, which includes parameters related to muscle strength, contraction characteristics and muscle deformation; simulates muscle movement deformation according to a virtual action optimization sequence and biomechanical parameters; calculates and applies muscle deformation on a skin mesh of a virtual human body model by applying mathematical models and algorithms; and generates virtual action demonstration data with muscle deformation according to the model's movement and muscle deformation.

[0105] The present invention can create a virtual human body model by importing a rigid body chain model and a parameterized muscle model into a virtual reality engine. This model can be used as a basis for subsequent operations such as action driving, skin mesh generation and action simulation. The virtual human body model can perform various interactions and operations in a virtual reality environment, and has more realistic and controllable characteristics. By mapping expert action data to a virtual human body model and driving the model for action, the movement of the virtual human body can be realized. This step can enable the virtual human body to accurately simulate according to the expert action data, and can respond to the user's interaction to achieve more natural and realistic action performance. By generating a high-fidelity skin mesh for the action-driven human body model, a realistic appearance can be added to the virtual human body model. This step can make the appearance of the virtual human body more realistic and detailed, and make the virtual human body more realistically presented in a virtual reality environment. Through skeletal animation technology, preliminary human body movements can be realized according to the virtual human body action model. This step can enable the virtual human body model to perform basic animation performance according to the input action data, and realize a simple action sequence. By smoothing the details of the virtual action sequence and biomechanical parameters, the movement of the virtual human body can be made smoother and more natural; this step can take into account the constraints of biomechanics and optimize the virtual action sequence, so that the movement of the virtual human body is more realistic and in line with the biomechanical laws of human movement. By simulating the muscle movement deformation of the virtual action optimization sequence based on biological parameters, the muscles of the virtual human body can show realistic deformation during the movement; this step can consider the biomechanical characteristics of human muscles and add muscle deformation to the virtual human body, making the virtual action demonstration more realistic and vivid. In summary, the effects of the above steps include: creating a virtual human model, achieving accurate action simulation, adding realistic skin appearance, achieving basic action sequences, optimizing action details, and simulating muscle deformation; these effects make the performance of the virtual human body in the virtual reality environment more realistic, natural and vivid, providing users with a more immersive interactive experience; in addition, these steps are also helpful for studying the human movement mechanism, optimizing sports training and rehabilitation programs, and conducting related research and applications in the field of virtual reality.

[0106] Preferably, step S4 comprises the following steps:

[0107] Step S41: Obtaining student action video data;

[0108] The embodiment of the present invention collects the sports video data of the trainees, which can be recorded by a camera or a mobile device; the quality and clarity of the video are ensured to facilitate subsequent analysis and processing.

[0109] Step S42: extracting action segments from the student's action video data, and annotating the action segments with action types and key frames, thereby obtaining an action annotation data set;

[0110] The embodiment of the present invention uses video editing software (such as Adobe Premiere Pro, Final Cut Pro, etc.) to edit the student's action video and extract the clips containing specific actions; based on the extracted action clips, each action is labeled, including the marking of the action type and key frames. The labeling can be performed manually or automatically with the help of computer vision algorithms.

[0111] Step S43: performing data enhancement on the action annotation data set, extracting a human skeleton model based on posture estimation, thereby obtaining skeleton action annotation data;

[0112] The embodiment of the present invention performs data enhancement on the action annotation dataset, generates more diverse data samples by scaling, rotating, translating and other operations on the data, and enhances the robustness of the training model; uses a posture estimation algorithm (such as OpenPose, DeepPose, etc.) to process the enhanced data and extracts key node information of the human skeleton model, such as joint positions and angles.

[0113] Step S44: performing skeleton-based motion alignment according to the skeleton motion annotation data and the expert motion data, thereby obtaining aligned student motion data;

[0114] The embodiment of the present invention aligns the skeleton motion annotation data with the expert motion data, and finds the correspondence between the student motion and the expert motion by comparing the positions and motion trajectories of key nodes; using the aligned student motion data, subsequent analysis and comparison can be performed more accurately.

[0115] Step S45: performing key joint feature enhancement expression on the aligned student action data, thereby obtaining student action feature data;

[0116] The embodiment of the present invention extracts feature information of key joint points, such as joint angles, speeds, accelerations, etc., based on the aligned student motion data; feature extraction algorithms (such as principal component analysis, convolutional neural networks, etc.) can be used to enhance the expression of features to reduce redundant information and highlight key features.

[0117] Step S46: Perform exercise movement recognition based on the virtual movement demonstration data and the student's movement feature data to obtain the student's movement pattern data.

[0118] The embodiment of the present invention uses machine learning algorithms (such as support vector machines, deep neural networks, etc.) to recognize exercise movements based on student movement feature data and virtual movement demonstration data; when training the model, the virtual movement demonstration data is used as positive samples and the student movement feature data is used as negative samples, and the recognition of the student's movement pattern is achieved through model training and classification.

[0119] The present invention can obtain the actual performance of the students under specific actions by acquiring the video data of the students' actions; these video data provide the basis for analyzing and evaluating the students' actions, and provide input data for subsequent steps. By intercepting and marking the video data of the students' actions, the complete video can be divided into action segments, and the action type and key frame are marked for each segment; the effect of this step is that the student's action data is effectively segmented and marked, providing an accurate data set for subsequent action analysis and recognition. By performing data enhancement and posture estimation on the action annotation data set, the action information of the human skeleton model can be extracted from the video data; the effect of this step is that the student's action data is converted into more abstract and processable skeleton action annotation data, providing a basis for subsequent action alignment and feature extraction. By aligning the student's skeleton action data with the expert's action data, the student's action can be compared and matched with the expert's standard; the effect of this step is to eliminate the differences in the student's actions, so that the student's action data is closer to the expert's action standard, providing a consistent benchmark for subsequent analysis and evaluation. By extracting and strengthening the key joint features of the aligned student motion data, important feature information can be extracted from the motion data; the effect of this step is to convert the student motion data into more compact and representative feature data, which facilitates subsequent motion recognition and analysis. By comparing and matching the virtual motion demonstration data with the student motion feature data, the student's motion pattern and performance can be identified; the effect of this step is to classify and identify the student's motion, provide personalized feedback and guidance to the student, and help them improve and optimize their motion skills. In summary, the effects of the above steps include: obtaining student motion data, dividing and annotating motion fragments, extracting skeletal motion annotation data, achieving motion alignment, extracting student motion features, and identifying student motion patterns; these effects make the student's motion data more processable and analyzable, which helps to deeply understand the student's motion performance, provide personalized feedback and guidance, and improve teaching and training effects; these steps provide a basis for subsequent motion analysis, evaluation and optimization, and help improve the student's motion skills and athletic ability.

[0120] Preferably, step S46 includes the following steps:

[0121] Step S461: decomposing the actions in the virtual action demonstration data into semantic action units, thereby obtaining a motion semantic data set;

[0122] The embodiment of the present invention analyzes and observes the virtual action demonstration data, decomposes each action into semantic action units, that is, the basic motion elements in the action; according to the semantic information of the action, a corresponding label or code is assigned to each semantic action unit to construct a motion semantic data set.

[0123] Step S462: extracting features based on convolution of kernels of different sizes on the trainee's motion feature data, and splicing feature channels at different scales to obtain multi-scale feature representation data, wherein the kernels of different sizes include a small-size kernel to capture local spatiotemporal details and a large-size kernel to capture large-scale motion patterns;

[0124] The embodiment of the present invention uses a convolutional neural network (CNN) to extract features from student motion feature data, uses convolution kernels of different sizes to perform convolution operations, small-size kernels are used to capture local spatiotemporal details, and large-size kernels are used to capture large-scale motion patterns, and feature channels at different scales are spliced ​​to obtain multi-scale feature representation data.

[0125] Step S463: constructing a spatiotemporal attention module based on the multi-scale feature representation data, and calculating the correlation weights between the spatiotemporal position and all other positions through the spatiotemporal attention module, thereby obtaining spatiotemporal dependent feature data, wherein the correlation weight calculation includes a temporal attention weight calculation based on capturing the temporal dependency of the action, and a spatial attention weight calculation based on capturing the correlation between the trunk and the limbs;

[0126] An embodiment of the present invention constructs a spatiotemporal attention module, which is used to calculate the correlation weights between spatiotemporal positions; the module is used to process multi-scale feature representation data to calculate the correlation weights between each spatiotemporal position and other positions; the calculation of the correlation weights includes the calculation of the temporal attention weights based on timing dependency and the calculation of the spatial attention weights based on the correlation between the trunk and limbs.

[0127] Step S464: annotating the expert action data according to the motion semantic data set, and training a preset LSTM model based on the annotated data set, thereby obtaining an action recognition model;

[0128] The embodiment of the present invention uses a motion semantic dataset to annotate expert motion data, and corresponds the semantic label of each motion with the expert motion data; constructs a long short-term memory network (LSTM) model, and uses the annotated dataset to train the model so that it can learn to recognize patterns of different motions.

[0129] Step S465: Input the spatiotemporal dependent feature data into the motion recognition model and perform exercise motion recognition to obtain the student's motion pattern data.

[0130] The embodiment of the present invention inputs the data processed with time-space dependent features into a trained motion recognition model, uses the motion recognition model to recognize and classify the student's motion pattern, and obtains the student's motion pattern data.

[0131] The present invention can decompose the virtual action demonstration data into semantic action units, so as to split the complex action into smaller semantic units and extract the basic components of the action; the effect of this step is to obtain a motion semantic data set, in which each semantic action unit represents a specific action fragment, which provides a basis for subsequent feature extraction and action recognition. By performing multi-scale feature extraction on the student action feature data, the details and overall pattern of the action can be captured from different scales; using convolution kernels of different sizes for feature extraction and splicing feature channels at different scales can obtain a more comprehensive and rich feature representation; the effect of this step is to improve the expressive ability of the feature, which helps to better capture the spatiotemporal characteristics of the student's action. By constructing a spatiotemporal attention module, the spatiotemporal dependency of the multi-scale feature representation data can be modeled and captured. The spatiotemporal attention module weights the important spatiotemporal positions and associated features by calculating the correlation weights of each spatiotemporal position with other positions. The effect of this step is to strengthen the temporal dependency and spatial correlation of the student's action features, thereby improving the accuracy and robustness of action recognition. By annotating the motion semantic data set and using the annotated data set to train the LSTM model, a model for motion recognition can be established; the effect of this step is that the model can learn and recognize the feature representations of different semantic action units, providing a reliable model foundation for subsequent motion recognition. By inputting the data processed with spatiotemporal dependent features into the motion recognition model, the recognition and classification of the student's movements can be achieved; the effect of this step is to match the student's movements with known exercises and obtain the student's motion pattern data, providing a basis for further analysis and evaluation; through this step, the student's movements can be accurately recognized and classified, thereby providing personalized and targeted guidance and feedback for the student's sports training.

[0132] Preferably, step S5 comprises the following steps:

[0133] Step S51: Detecting key points of the trainee's movements and constructing a skeleton model according to the trainee's movement video data, thereby obtaining the trainee's biological parameters;

[0134] The embodiments of the present invention use computer vision technology, such as posture estimation or human key point detection methods, to process video data of student movements; extract student biological parameters, such as joint angles, body postures, etc., by analyzing the human posture and joint positions in the video; construct a student's skeletal model based on the extracted joint position and posture information for subsequent biomechanical analysis and modeling. OpenPose: A commonly used human posture estimation library for detecting human key points; Kinect or camera: used to collect video data of student movements; Computer vision algorithms and deep learning models: used for human key point detection and posture estimation.

[0135] Step S52: performing virtual and real action difference measurement analysis on the student's motion pattern data according to the virtual action demonstration data to obtain action difference measurement data;

[0136] The embodiment of the present invention uses virtual action demonstration data as a reference to compare and analyze the student's motion pattern data to measure the difference between the student and the virtual action; based on the difference measurement index (such as Euclidean distance, angle difference, etc.), the student's action difference measurement data is calculated to reflect the degree of difference between the student and the virtual action. Motion capture system: used to obtain virtual action demonstration data and student motion pattern data; motion difference measurement algorithm: according to specific needs and data characteristics, select an appropriate measurement method to measure the action difference.

[0137] Step S53: constructing a personalized rigid body-muscle model representation according to the student's biological parameters and biomechanical parameters, thereby obtaining a personalized biomechanical model;

[0138] The embodiment of the present invention uses the biological parameters and motion data of the trainees to construct a rigid body model, including the skeletal structure and joint connection relationship; based on the biomechanical parameters of the trainees, such as muscle length, strength, etc., combined with the rigid body model, a personalized rigid body-muscle model representation is constructed to simulate the trainees' motion process. Rigid body model construction software: such as 3D modeling software such as Blender and Maya; biomechanical parameter database: used to obtain the biomechanical parameters of the trainees; biomechanical simulation tools: such as OpenSim and other tools for constructing and simulating rigid body-muscle models.

[0139] Step S54: performing constraint optimization according to the personalized biomechanical model and the action difference measurement data, thereby obtaining action correction solution data;

[0140] The embodiment of the present invention combines the personalized biomechanical model with the action difference measurement data to establish an optimization problem; through the constraint optimization method, the action correction scheme is found to make the trainee's movement pattern closer to the virtual action; during the optimization process, various constraints can be set, such as joint angle range, muscle strength limit, etc., to ensure that the generated action correction scheme meets physiological constraints and individual characteristics. Optimization algorithm: such as nonlinear programming, evolutionary algorithm, etc.; optimization library or framework: such as SciPy, MATLAB, etc.; constraint setting: set appropriate constraints according to biomechanical knowledge and actual needs.

[0141] Step S55: evaluating the difference between the movements before and after correction according to the movement correction scheme data, and thereby obtaining training evaluation report data;

[0142] The embodiment of the present invention corrects the trainee's motion pattern according to the motion correction scheme, uses the corrected motion data to compare with the virtual motion, calculates the motion difference measurement data before and after the correction, and evaluates the correction effect; generates training evaluation report data according to the evaluation results, including information such as the degree of improvement and evaluation indicators. Motion capture system: used to obtain the corrected trainee motion data; motion difference measurement algorithm: the same as the method in step S52, used to calculate the motion difference measurement data; data analysis and report generation tools: such as Python, MATLAB, etc., used to analyze data and generate evaluation reports.

[0143] Step S56: constructing a training evaluation decision model based on a decision tree model according to the training evaluation report data.

[0144] The embodiment of the present invention performs data analysis and feature extraction based on the training evaluation report data; constructs a training evaluation decision model based on a decision tree model or other machine learning algorithms for automated evaluation and decision-making; trains the model and performs validation and tuning to ensure the accuracy and reliability of the model. Data analysis and feature extraction tools: such as Python's data analysis library (such as NumPy, Pandas), feature selection algorithms, etc.; machine learning libraries and algorithms: such as scikit-learn, XGBoost, etc., for building decision models; model validation and tuning methods: such as cross-validation, grid search, etc.

[0145] The present invention can obtain the biological parameters of the students, including joint positions, posture angles and other information, by detecting the key points of the human body and constructing a skeleton model for the students' action video data; the effect of this step is to provide the biological characteristics and posture information of the students during the movement process, and provide a basis for the subsequent personalized biomechanical model construction and action analysis. By comparing and measuring the students' movement pattern data with the virtual action demonstration data, the difference between the students' movements and the standard movements can be evaluated; the effect of this step is to obtain action difference measurement data, which can be used to evaluate the students' sports skill level and the accuracy of action execution. By combining the students' biological parameters and biomechanical parameters, a personalized rigid body-muscle model can be constructed to simulate the students' biomechanical characteristics and action execution ability; the effect of this step is to obtain a personalized biomechanical model, which can more accurately describe the students' movement characteristics and action execution ability. By combining the personalized biomechanical model with the action difference measurement data, constraint optimization can be performed to find the optimal action correction plan; the effect of this step is to obtain action correction plan data, which can guide students to improve and optimize their actions and improve the accuracy and effect of their actions. By comparing the differences in movements before and after correction, the effect of the movement correction and the degree of improvement can be evaluated; the effect of this step is to obtain training evaluation report data, which can provide a quantitative evaluation of the student's movement improvement, help understand the effectiveness of the correction plan and guide the improvement strategy. By building a decision tree model based on the training evaluation report data, the evaluation results can be associated with the corresponding decisions; the effect of this step is to obtain a training evaluation decision model, which can make corresponding decisions based on the student's training evaluation results, such as formulating personalized training plans, adjusting movement correction strategies, etc., to achieve more effective training and improve the student's sports performance.

[0146] Preferably, step S52 includes the following steps:

[0147] Step S521: aligning the student's motion pattern data with the virtual motion demonstration data, and using a dynamic time warping algorithm to calculate the similarity of the skeletal motion trajectory, thereby obtaining skeletal trajectory comparison data;

[0148] The embodiment of the present invention time-aligns the student's motion pattern data and virtual action demonstration data to ensure that the length and frequency of the motion sequences of the two are consistent. The dynamic time warping (DTW) algorithm is used to calculate the similarity of the skeletal motion trajectory; the DTW algorithm can find the best matching path between the two time series while considering time delay and deformation. According to the similarity calculated by the DTW algorithm, the skeletal trajectory comparison data is obtained to reflect the degree of difference in trajectory between the student's motion pattern and the virtual action. Motion capture system: used to obtain the student's motion pattern data and virtual action demonstration data; time alignment method: interpolation or time warping and other technologies can be used to ensure that the length and frequency of the data are consistent; dynamic time warping algorithm: commonly used algorithms include DTW algorithm based on distance measurement and DTW algorithm based on dynamic programming; distance measurement method: Euclidean distance, Manhattan distance, etc. can be used to calculate the similarity between skeletal motion trajectories.

[0149] Step S522: Calculating the difference of the biological parameters of joint angle, torque and tendon transfer ratio according to the biological parameters and biomechanical parameters of the trainee, thereby obtaining biological parameter difference data;

[0150] The embodiment of the present invention uses the biological parameters and biomechanical parameters of the trainee to analyze the values ​​of biological parameters such as joint angle, torque and tendon transfer ratio; compares the biological parameters of the trainee with the corresponding parameters of the virtual action demonstration data, calculates the difference value, and reflects the degree of difference between the biological parameters of the trainee and the virtual action. Biomechanical parameter database: used to obtain the biomechanical parameters of the trainee; numerical calculation tools: such as MATLAB, Python numerical calculation library, used to calculate parameters such as joint angle, torque and tendon transfer ratio; parameter comparison and difference calculation method: according to the nature of the specific parameters, select an appropriate comparison method, such as absolute difference, relative difference, etc.

[0151] Step S523: merging the skeleton trajectory comparison data and the biological parameter difference data into motion difference measurement data.

[0152] The embodiment of the present invention combines the bone trajectory comparison data and the biological parameter difference data to form comprehensive motion difference measurement data; different data can be weighted or normalized according to specific needs to obtain a unified measurement index. Data processing tools: such as Python, MATLAB, etc., are used to process and merge data; data weighting and normalization methods: according to the importance or range difference of the data, weighted average, maximum and minimum normalization, etc. can be used to ensure the comparability and consistency of the data.

[0153] The present invention can quantitatively compare the similarity between the actual movement of the student and the standard movement by aligning the student's motion pattern data with the virtual action demonstration data and using the dynamic time warping algorithm to calculate the similarity of the skeletal motion trajectory; the effect of this step is to obtain the skeletal trajectory comparison data, which can evaluate whether the student's action execution is consistent with the standard action, and provide a basis for the subsequent action difference measurement; by comparing the differences between the student's biological parameters and biomechanical parameters and the standard parameters, including joint angles, torques, and tendon transfer ratios, the degree of difference between the student's biomechanical characteristics and the standard model can be evaluated; the effect of this step is to obtain biological parameter difference data, which can quantitatively understand the difference between the student's biomechanical characteristics and the standard model. By merging the skeletal trajectory comparison data and the biological parameter difference data, comprehensive action difference measurement data can be obtained; the effect of this step is to comprehensively consider the student's motion trajectory similarity and biological parameter differences, and provide comprehensive action difference measurement information; these data can be used to evaluate the student's action performance, guide action correction and improvement, and provide a basis for personalized training plans and decisions.

[0154] Preferably, step S56 includes the following steps:

[0155] Step S561: formatting and cleaning the training evaluation report data, and extracting quantitative indicators and text to obtain text feature data and quantitative indicator data;

[0156] The embodiment of the present invention formats and cleans the training evaluation report data, removes unnecessary punctuation marks, blank characters, etc., and ensures the consistency and standardization of the data; extracts the quantitative indicator data therein according to the content of the evaluation report, which involves using regular expressions or specific text processing methods to identify and extract relevant information such as numbers, percentages, and time; similarly, extracts text feature data from the evaluation report. This includes extracting keywords, phrases, descriptive texts, etc.; and performs appropriate preprocessing on the extracted quantitative indicator data and text feature data, such as removing stop words and performing stemming processing.

[0157] Step S562: normalizing the quantitative index data to obtain normalized quantitative vector data; performing word segmentation and word vector encoding on the text feature data to obtain distributed text feature data;

[0158] The embodiment of the present invention normalizes the extracted quantitative indicator data and maps it to a uniform numerical range. Common normalization methods include maximum and minimum normalization, Z-score normalization, etc.; segmentation is performed on the text feature data to split the text into a sequence of words or phrases; word vector encoding is performed on the text feature data after segmentation to convert each word or phrase into a vector representation. Common word vector encoding methods include bag of words model, TF-IDF, Word2Vec, GloVe, etc.

[0159] Step S563: fusing feature information of different modalities according to the normalized quantized vector data and the distributed text feature data, and constructing a composite feature group, thereby obtaining composite feature data;

[0160] The embodiment of the present invention performs feature fusion on normalized quantized vector data and distributed text feature data, and different methods can be used to combine the two feature information. Common feature fusion methods include concatenation, averaging, weighted averaging, etc. According to actual needs, a suitable feature fusion strategy can be selected to organically combine feature information of different modalities to form composite feature data.

[0161] Step S564: using the composite feature data as a training set for a decision tree model and performing parameter tuning to obtain a preliminary decision tree model;

[0162] The embodiment of the present invention uses the constructed composite feature data as input, divides it into a training set and a validation set; uses a decision tree algorithm (such as CART, ID3, C4.5, etc.) to build a preliminary decision tree model. The decision tree model class in the scikit-learn library in Python can be used for modeling; the decision tree model is parameter tuned, and the appropriate parameter settings are selected through methods such as cross-validation, and grid search, random search, etc. can be used for parameter search and selection.

[0163] Step S565: Perform a generalization performance evaluation on the preliminary decision tree model on the retained test set, and perform structural optimization on the preliminary decision tree model based on the evaluation results, so as to obtain a training evaluation decision model.

[0164] The embodiment of the present invention uses the reserved test set to evaluate the performance of the preliminary decision tree model and evaluate the generalization ability of the model on new data. Common evaluation indicators include accuracy, precision, recall, F1 value, etc. The structure of the preliminary decision tree model is optimized according to the evaluation results. Pruning operations can be considered to reduce the risk of overfitting and improve the generalization performance of the model. Iterative optimization steps S564 and S565 are performed until a satisfactory training evaluation decision model is obtained.

[0165] The present invention can obtain structured data, including quantitative indicators and text features, by formatting, cleaning and extracting training evaluation report data; the effect of this step is to convert the original evaluation report data into a processable data form, in preparation for subsequent feature extraction and model construction. By normalizing the quantitative indicator data, the indicator data of different ranges and units can be converted into a unified scale, eliminating the dimension effect, and facilitating subsequent feature fusion and model training; at the same time, by segmenting and encoding the text feature data, the text information can be converted into a numerical feature representation, so that the text feature can be used together with other numerical features. By fusing the normalized quantitative vector data and the distributed text feature data, the feature information of different modes can be combined to provide a more comprehensive and diversified feature representation; the effect of this step is to obtain composite feature data, which contains features of different dimensions and types, and can better reflect the training evaluation of the trainees. By using the composite feature data as the training set of the decision tree model, the relationship between the features can be learned by the decision tree algorithm, and a preliminary decision tree model can be established; by parameter tuning, the performance and generalization ability of the model can be optimized, and the degree of fit of the model to the training evaluation data can be improved. By evaluating the preliminary decision tree model on the reserved test set, the generalization ability and prediction performance of the model can be evaluated. Based on the evaluation results, the decision tree model can be structurally optimized, including pruning, adjusting parameters, etc., to improve the accuracy and stability of the model; the effect of this step is to obtain a training evaluation decision model, which can make corresponding decisions and recommendations based on the characteristic data of the trainees, and help formulate personalized training plans and improvement strategies.

[0166] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0167] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for constructing a fitness training evaluation decision model based on VR virtual digital human, It is characterized in that The following steps are involved: Step S1: capturing the expert's practice movements through a high-precision inertial camera group network to obtain expert movement data, wherein the expert movement data includes joint movement angle data and muscle activity potential data; Step S2: Analyze the joint torque, tendon transfer ratio and inertial force of the expert action data based on the rigid body dynamics model and musculoskeletal simulation to obtain biomechanical parameters; Step S3: importing the expert motion data and biomechanical parameters into the virtual reality engine, and performing virtual digital human motion demonstration generation through skeletal drive and inverse kinematics to obtain virtual motion demonstration data; Step S4: using computer vision technology to enhance the feature expression of key joints of the pre-acquired student action video data, and performing exercise action recognition on the student action video data according to the virtual action demonstration data, thereby obtaining the student movement pattern data; Step S5: Perform human biological parameter analysis on the student's action video data to obtain the student's biological parameters; perform difference measurement analysis on the student's motion pattern data based on the virtual action demonstration data to obtain action difference measurement data; perform action adjustment based on the biomechanical model according to the student's biological parameters and the action difference measurement data, and generate training evaluation report data; construct a training evaluation decision model based on the decision tree model according to the training evaluation report data.

2. According to the method for constructing a fitness training evaluation decision model based on VR virtual digital human according to claim 1, It is characterized in that Step S1 includes the following steps: Step S11: using texture-free blue-green as the background, arranging multiple high-precision inertial capture cameras to cover the full view, and obtaining a high-precision inertial camera group network; Step S12: Paste reflective markers on key parts of the expert's body and wear a surface electromyography sensor. The expert performs a complete demonstration of the exercise action process in the inertial camera group network to collect the original electromyography data and original 3D position data of the markers; Step S13: removing noise and missing information from the original electromyographic data and the original 3D position data, and synchronizing the timestamps of the position and electromyographic data, thereby obtaining muscle activity data and 3D marker position data; Step S14: constructing a skeleton model according to the muscle activity data and the 3D marker point position data by using biomechanical modeling software, thereby obtaining expert skeleton model data; Step S15: Bind the 3D marker point position data to the expert skeleton model data to calculate the joint motion angle, thereby obtaining the joint motion angle data; estimate the muscle activity level and action intensity according to the muscle activity data, thereby obtaining the muscle action potential data; Step S16: Store the joint motion angle data and muscle action potential data as expert motion data.

3. The method for constructing a fitness training evaluation decision model based on VR virtual digital human according to claim 2, It is characterized in that Step S2 includes the following steps: Step S21: constructing a rigid body chain model and a muscle model according to the expert action data; mapping the muscle activity potential data into the activity parameters of the muscle model, thereby obtaining a parameterized muscle model; Step S22: using the joint motion angle data as motion input to drive the rigid body chain model to move, and solving the force and torque of each joint based on Newton's second law to obtain the torque value of each joint at each time step, thereby obtaining the joint torque data; Step S23: using the joint torque data as input, inversely solving the muscle contraction force of the parameterized muscle model, thereby obtaining the muscle contraction force data; Step S24: Calculate the moment acting on the bone based on the position of the muscle attachment point according to the muscle contraction force data, so as to obtain the bone moment data; calculate the tendon transfer ratio data according to the bone moment data and the muscle contraction force data, wherein the tendon transfer ratio = bone moment / muscle contraction force; Step S25: decomposing the joint motion angle data into the motion of each rigid body according to the rigid body chain model, and calculating the linear acceleration and angular acceleration based on the rigid body, thereby obtaining the rigid body motion parameter data; Step S26: deriving the inertial force and equilibrium moment of each body part based on Newton's law of motion from the rigid body motion parameter data, thereby obtaining inertial force data; Step S27: Perform bio-parameter analysis based on the joint torque data, tendon transfer ratio data and inertial force data to obtain biomechanical parameters.

4. The method for constructing a fitness training evaluation decision model based on VR virtual digital human according to claim 3, It is characterized in that Step S27 includes the following steps: Step S271: Integrate the joint torque data, tendon transfer ratio data and inertial force data into an initial biomechanical parameter set; Step S272: Calculate the muscle contraction work according to the joint torque data to obtain muscle contraction work data; calculate the loss work done by the inertial force according to the inertial force data to obtain inertial loss work data; Step S273: Calculate the total energy during the action according to the muscle contraction work data and the inertia loss work data to obtain energy consumption data, wherein energy consumption = muscle contraction work + inertia loss work; Step S274: Integrate the joint torque data, tendon transfer ratio data, inertial force data and energy consumption data into biomechanical parameters.

5. According to the method for constructing a fitness training evaluation decision model based on VR virtual digital human according to claim 4, It is characterized in that Step S3 includes the following steps: Step S31: importing the rigid body chain model and the parameterized muscle model into a virtual reality engine to obtain a virtual human body model; Step S32: performing data mapping on the virtual human body model according to the expert motion data, and performing motion driving on the human body model, thereby obtaining a motion driven human body model; Step S33: generating a high-fidelity skin mesh model based on bones for the action-driven human body model, thereby obtaining a virtual human action model with a skin mesh; Step S34: preliminarily realizing human body movement according to the virtual human body action model through skeletal animation technology, thereby obtaining a virtual action sequence; Step S35: performing detailed motion smoothing based on inverse kinematics constraints according to the virtual motion sequence and biomechanical parameters, thereby obtaining a virtual motion optimization sequence; Step S36: Perform muscle movement deformation simulation based on biological parameters on the virtual action optimization sequence, so as to obtain virtual action demonstration data with muscle deformation.

6. The method for constructing a fitness training evaluation decision model based on VR virtual digital human according to claim 5, It is characterized in that Step S4 includes the following steps: Step S41: Obtaining student action video data; Step S42: extracting action segments from the student's action video data, and annotating the action segments with action types and key frames, thereby obtaining an action annotation data set; Step S43: performing data enhancement on the action annotation data set, extracting a human skeleton model based on posture estimation, thereby obtaining skeleton action annotation data; Step S44: performing skeleton-based motion alignment according to the skeleton motion annotation data and the expert motion data, thereby obtaining aligned student motion data; Step S45: performing key joint feature enhancement expression on the aligned student action data, thereby obtaining student action feature data; Step S46: Perform exercise movement recognition based on the virtual movement demonstration data and the student's movement feature data to obtain the student's movement pattern data.

7. The method for constructing a fitness training evaluation decision model based on VR virtual digital human according to claim 6, It is characterized in that Step S46 includes the following steps: Step S461: decomposing the actions in the virtual action demonstration data into semantic action units, thereby obtaining a motion semantic data set; Step S462: extracting features based on convolution of kernels of different sizes on the trainee's motion feature data, and splicing feature channels at different scales to obtain multi-scale feature representation data, wherein the kernels of different sizes include a small-size kernel to capture local spatiotemporal details and a large-size kernel to capture large-scale motion patterns; Step S463: constructing a spatiotemporal attention module based on the multi-scale feature representation data, and calculating the correlation weights between the spatiotemporal position and all other positions through the spatiotemporal attention module, thereby obtaining spatiotemporal dependent feature data, wherein the correlation weight calculation includes a temporal attention weight calculation based on capturing the temporal dependency of the action, and a spatial attention weight calculation based on capturing the correlation between the trunk and the limbs; Step S464: annotating the expert action data according to the motion semantic data set, and training a preset LSTM model based on the annotated data set, thereby obtaining an action recognition model; Step S465: Input the spatiotemporal dependent feature data into the motion recognition model and perform exercise motion recognition to obtain the student's motion pattern data.

8. The method for constructing a fitness training evaluation decision model based on VR virtual digital human according to claim 7, It is characterized in that Step S5 includes the following steps: Step S51: Detecting key points of the trainee's movements and constructing a skeleton model according to the trainee's movement video data, thereby obtaining the trainee's biological parameters; Step S52: performing virtual and real action difference measurement analysis on the student's motion pattern data according to the virtual action demonstration data to obtain action difference measurement data; Step S53: constructing a personalized rigid body-muscle model representation according to the student's biological parameters and biomechanical parameters, thereby obtaining a personalized biomechanical model; Step S54: performing constraint optimization according to the personalized biomechanical model and the action difference measurement data, thereby obtaining action correction solution data; Step S55: evaluating the difference between the movements before and after correction according to the movement correction scheme data, and thereby obtaining training evaluation report data; Step S56: constructing a training evaluation decision model based on a decision tree model according to the training evaluation report data.

9. The method for constructing a fitness training evaluation decision model based on VR virtual digital human according to claim 8, It is characterized in that Step S52 includes the following steps: Step S521: aligning the student's motion pattern data with the virtual action demonstration data, and using a dynamic time warping algorithm to calculate the similarity of the skeletal motion trajectory, thereby obtaining skeletal trajectory comparison data; Step S522: Calculating the difference of the biological parameters of joint angle, torque and tendon transfer ratio according to the biological parameters and biomechanical parameters of the trainee, thereby obtaining biological parameter difference data; Step S523: merging the skeleton trajectory comparison data and the biological parameter difference data into motion difference measurement data.

10. The method for constructing a fitness training evaluation decision model based on VR virtual digital human according to claim 9, It is characterized in that Step S56 includes the following steps: Step S561: formatting and cleaning the training evaluation report data, and extracting quantitative indicators and text to obtain text feature data and quantitative indicator data; Step S562: normalizing the quantitative index data to obtain normalized quantitative vector data; performing word segmentation and word vector encoding on the text feature data to obtain distributed text feature data; Step S563: fusing feature information of different modalities according to the normalized quantized vector data and the distributed text feature data, and constructing a composite feature group, thereby obtaining composite feature data; Step S564: using the composite feature data as a training set for a decision tree model and performing parameter tuning to obtain a preliminary decision tree model; Step S565: Perform a generalization performance evaluation on the preliminary decision tree model on the retained test set, and perform structural optimization on the preliminary decision tree model based on the evaluation results, so as to obtain a training evaluation decision model.

Citation Information

Patent Citations

  • Muscle strength model optimizing method for humanoid robot synergic movement

    CN103761392A

  • Body teaching method and body teaching system based on virtual reality

    CN106097787A

  • Interaction method and system of digital twin virtual human and human body, and electronic equipment

    CN115543094A

  • Vision Based Human Activity Recognition and Monitoring System for Guided Virtual Rehabilitation

    US20110054870A1

Cited By

  • Virtual reality interaction method, electronic device, storage medium and program product

    CN119379958A

  • Fitness guidance system and method based on AI attitude analysis and MR man-machine interaction

    CN119649997A

  • Fitness guidance system and method based on AI posture analysis and MR human-computer interaction

    CN119649997B

  • Interactive physical education method and system based on augmented reality driving

    CN120088108A

  • Personalized training plan generation method and system in element universe smart sports

    CN120126680A