Sports training method, device and system based on visual AI model
Through the motion training method based on visual AI model, users' motion data are collected and analyzed in real time, and personalized guidance is provided, which solves the problem of long-term tracking and guidance of traditional sports training, improves the exercise efficiency and standardization, and is suitable for the scarcity of medical resources. area.
Patent Information
- Application Number
- CN202411074986.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-07
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-08-07
AI Technical Summary
Traditional sports training requires long-term tracking and guidance, and it is difficult for users to obtain professional guidance in areas with scarce medical resources, resulting in unstable sports movement standards and affecting the recovery of sports functions.
Using a motion training method based on visual AI model, users' information is collected through mobile terminals, joint angle data is collected in real time, motion curves are drawn, motion progress information is calculated and corrected, and personalized and intelligent motion training guidance is provided.
Personalized exercise training at home or outside the hospital has been achieved, which has improved exercise efficiency and participation, reduced dependence on medical staff, and ensured the standardization and effectiveness of exercise training.
Smart Images

Figure CN118588239B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of data processing technology, and in particular relates to a motion training method, device and system based on a visual AI model. Background Art
[0002] As the country promotes the scientific and technological revolution and the industrial revolution, we should effectively track the development and application of artificial intelligence in my country, deepen research on important issues such as promoting basic research and original innovation in the field of artificial intelligence, deep integration of artificial intelligence and industrial development, in-depth application of artificial intelligence in areas such as people's livelihood security and social governance, artificial intelligence governance and international cooperation, and strive to put forward targeted and operational countermeasures and suggestions to help artificial intelligence better empower the construction of a modern industrial system and high-quality development, and better create and serve the people's better life.
[0003] We will use the new generation of artificial intelligence as a driving force to promote leapfrog development of science and technology, optimization and upgrading of industries, and overall leap in productivity. We will promote new breakthroughs in my country's artificial intelligence technological innovation, rapid development of the artificial intelligence industry, and in-depth expansion of its integration and application with modern industries. We will take solid steps in cultivating new quality productivity, building a modern industrial system, and promoting high-quality development.
[0004] In the process of rapid development of the new generation of artificial intelligence industry and in-depth expansion of its integration and application with modern industries, people are paying more and more attention to the importance of combining artificial intelligence with sports. Skeletal movement is actually the movement of skeletal muscles responsible for bones. Exercise is the action of certain specific skeletal muscles. Users may face challenges in restoring or improving their motor functions due to various factors, including diseases, accidental injuries, etc.
[0005] Traditional sports training often requires long-term tracking and guidance in order to correct its sports movements and exercise frequency, etc., which is a challenge to the instructor's time and energy. At the same time, medical resources in some areas are scarce, and users cannot get professional sports guidance in time, which further increases the difficulty of exercise. In addition, in the current cognition, the standards for sports movements are relatively broad, and most of them are summaries of the experience of professionals. This will inevitably lead to certain movements that are not stable enough to help restore the function of specific skeletal muscles, making it impossible to enjoy high-quality sports training.
[0006] However, with the continuous development of computer vision and deep learning technology, we have the ability to realize a medical service system that is child-centered and family-centered. Based on this point of view, we hope to provide a convenient and effective exercise training method for exercise programs. Combined with mobile devices such as smartphones, these technologies can be used to provide personalized, intelligent, and professional exercise training programs at home or outside the hospital, and realize comprehensive monitoring and management of the exercise process. This way of exercise can not only alleviate the problem of shortage of medical resources, but also improve exercise efficiency and participation in exercise training, providing strong support for the healthy growth and future development of users. Summary of the invention
[0007] The present invention provides a motion training method, device and system based on a visual AI model, aiming to solve the above-mentioned technical problems.
[0008] The present invention is implemented in this way. On the one hand, the present application provides a motion training method based on a visual AI model, comprising:
[0009] Collect user information of users to be trained through mobile terminals;
[0010] Acquire the user's training data information according to the user information, wherein the training data information includes the user's exercise training times, current planned exercise progress information and current exercise progress information;
[0011] Real-time collection of dynamic training data during the user's sports training process, wherein the dynamic training data includes the joint angles during the user's sports training process and conclusion data of the joint angles given by the visual AI model;
[0012] Draw an actual motion curve of the user according to the collected joint angles and conclusion data;
[0013] Calculate the actual exercise progress information of the user after the exercise training according to the actual exercise curve, the number of training times of the user, the current planned exercise progress information and the current exercise progress information;
[0014] The current planned exercise progress information is corrected according to the actual exercise progress information and the number of exercise trainings to obtain the corrected planned exercise progress information to guide the user to perform exercise training.
[0015] Preferably, drawing the actual motion curve of the user according to the collected joint angles and conclusion data comprises:
[0016] Obtaining functional recovery indicator data related to user information;
[0017] According to the statistical algorithm, the actual motion curve including the P value (with a statistical significance of less than 0.05) with the horizontal axis being time and the vertical axis being motion state is calculated based on the joint angle, conclusion data and functional recovery index data;
[0018] The time is in days, weeks or months; the functional recovery index data at least includes the recovery angle of the disabled joint, muscle strength, mobility, and imaging assessment results.
[0019] Preferably, obtaining the actual exercise progress information of the user after exercise training according to the actual exercise curve, the number of training times of the user, the current planned exercise progress information and the current exercise progress information includes:
[0020] Acquiring functional recovery index data of the user after the current exercise training according to the actual exercise curve;
[0021] Compare the user's functional recovery index data after the current exercise training with the dynamic training data in the current exercise progress information;
[0022] According to the comparison result, combined with the user's training times and the current planned exercise progress information, the actual exercise progress information of the user after exercise training is obtained.
[0023] Preferably, the current planned exercise progress information is corrected according to the actual exercise progress information and the number of exercise training times to obtain the corrected planned exercise progress information to guide the user to perform exercise training, including:
[0024] Obtain the planned exercise training times, planned exercise training time and planned exercise training indicator data in the current planned exercise progress information;
[0025] Acquire the activity angle of the disabled joint after actual sports training according to the actual sports progress information;
[0026] The planned exercise training times and the planned exercise training time are corrected according to the activity angle and the number of exercise training times, and new planned exercise training time, planned exercise training times and planned exercise training indicator data are obtained as corrected planned exercise progress information.
[0027] Preferably, the planned number of exercise training sessions and the planned exercise training time are modified according to the activity angle and the number of exercise training sessions, including:
[0028] Searching the preset rule database for the number of exercise training and the exercise training time corresponding to the activity angle and the number of exercise training respectively;
[0029] Compare the exercise training times and exercise training time with the planned exercise training times and planned exercise training time, and calculate the difference between the two respectively;
[0030] Add half of the difference between the planned exercise training times and the planned exercise training time, and then take the integer to get the new planned exercise training time and planned exercise training times.
[0031] Preferably, real-time acquisition of dynamic training data during the user's sports training process includes:
[0032] Obtain the joint angle of the user's exercise training and the time to maintain the joint angle in real time;
[0033] Data association is performed based on a combination of generative models and recognition capture;
[0034] Comparing the joint angles of the user's exercise training and the time the joint angles are maintained according to the data in the preset rule database, and making a judgment and analysis on the joint angles of the user's exercise training and the time the joint angles are maintained;
[0035] The joint angles that conform to the data in the preset rule database and the conclusion data given by the visual AI model on the joint angles are used as the dynamic training data to be collected during the user's motion training process.
[0036] Preferably, the method further comprises:
[0037] Acquire the training data of each exercise training of the user through the server, wherein the training data includes the angle, position, number of times and the number of qualified times of the exercise training of the disabled joint;
[0038] Classify the training data;
[0039] The classified training data, actual motion curve and actual motion progress information are trained to obtain key data affecting the motion training, and a corresponding preset rule database is established according to the key data.
[0040] Preferably, the classified training data, the actual motion curve and the actual motion progress information are trained, including:
[0041] Detecting disabled joints of the user to obtain disability information of the user's joints;
[0042] Obtaining a planned movement value by comparing the disability information with a preset rule database including at least a movement knowledge graph or standard movement actions;
[0043] According to the planned exercise values, the required exercise cycle and training content are calculated, and the angle of joint exercise training required by the user every day is obtained, and it is used as the key data affecting exercise training to guide the user to perform exercise training.
[0044] On the other hand, the present application also proposes a sports training device based on a visual AI model, and the sports training device based on a visual AI model includes:
[0045] An identity collection module, used to collect user information of the user to be trained;
[0046] An information acquisition module, used to acquire the user's training data information according to the user information;
[0047] The data acquisition module is used to collect dynamic training data of the user during the sports training process in real time;
[0048] The curve drawing module is used to draw the user's actual motion curve based on the collected joint angles and conclusion data;
[0049] An information calculation module, used to calculate the actual exercise progress information of the user after exercise training according to the actual exercise curve, the number of training times of the user, the current planned exercise progress information and the current exercise progress information;
[0050] The information correction module is used to correct the above-mentioned current planned exercise progress information according to the actual exercise progress information and the number of exercise trainings, and obtain the corrected planned exercise progress information to guide the user to perform exercise training.
[0051] On the other hand, the present application also provides a sports training system based on a visual AI model, including a mobile terminal and a server terminal, wherein:
[0052] The mobile terminal collects user information of the user to be trained based on the visual AI model, and the mobile terminal is also used to obtain the user's training data information according to the user information and to collect dynamic training data of the user during the training process in real time;
[0053] The server is used to draw the user's actual motion curve based on the collected joint angles and conclusion data. The server is also used to calculate the actual motion progress information of the user after exercise training based on the actual motion curve, the user's training times, the current planned motion progress information and the current motion progress information. The server is also used to correct the above-mentioned current planned motion progress information based on the actual motion progress information and the number of exercise trainings to obtain the corrected planned motion progress information to guide the user in exercise training.
[0054] The present invention provides a motion training method based on a visual AI model. Compared with the prior art, the present invention has the following beneficial effects: the motion training method based on a visual AI model comprises: collecting user information through a mobile terminal; obtaining the user's training data information according to the user information, the training data information comprising the user's motion training times, current planned motion progress information and current motion progress information; collecting dynamic training data during the user's motion training in real time, the dynamic training data comprising the joint angles during the user's motion training and conclusion data given for the joint angles by the visual AI model; drawing the user's actual motion curve according to the collected joint angles and conclusion data; calculating the user's actual motion progress information after the motion training according to the actual motion curve, the user's training times, the current planned motion progress information and the current motion progress information; correcting the above-mentioned current planned motion progress information according to the actual motion progress information and the motion training times, and obtaining the corrected planned motion progress information to guide the user to perform motion training. The motion training method based on the visual AI model provided by the present application is based on the training data information and dynamic training data collected by the mobile terminal based on the visual AI model, and then the actual motion progress information of the user after the motion training is obtained by corresponding calculation according to the training data information and the dynamic training data, and then the planned motion progress information is corrected by the actual motion information, so as to obtain the corrected new planned motion progress information to guide the user to perform motion training, so that the user can have a corresponding motion training guidance during the motion training process, and the method of the present application can effectively correct the subsequent motion training plan, so as to guide the subsequent user in the number of training times and the joint angle of the disabled joint during the motion training process, etc., and can effectively and completely guide and process the user's motion training, which can not only solve the problem of the user performing standard and effective motion training in the case of unlimited training locations and expert guidance. Through the motion training method based on the visual AI model of the present application, on the one hand, the user's motion training effect can be effectively improved, and on the other hand, the dependence and reliability of motion training on medical staff can be effectively reduced. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] Figure 1 is a structural block diagram of a sports training system based on a visual AI model provided by an embodiment of the present invention;
[0056] Figure 2 It is a functional module block diagram of a sports training device based on a visual AI model provided by an embodiment of the present invention;
[0057] Figure 3 It is a flowchart of a motion training method based on a visual AI model provided by an embodiment of the present invention;
[0058] Figure 4is a flow chart of another motion training method based on a visual AI model provided by an embodiment of the present invention;
[0059] Figure 5 is a flow chart of another motion training method based on a visual AI model provided by an embodiment of the present invention;
[0060] Figure 6 is a flow chart of another motion training method based on a visual AI model provided by an embodiment of the present invention;
[0061] Figure 7 is a flow chart of another motion training method based on a visual AI model provided by an embodiment of the present invention;
[0062] Figure 8 is a flow chart of another motion training method based on a visual AI model provided by an embodiment of the present invention;
[0063] Fig. 9 is a flow chart of another motion training method based on a visual AI model provided by an embodiment of the present invention;
[0064] Fig.10 It is a flowchart of another motion training method based on a visual AI model provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0065] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0066] like Figure 1 As shown in FIG. 1 , a structural block diagram of a motion training system based on a visual AI model provided by the present application is shown. The motion training methods based on a visual AI model disclosed in the present application are all processed based on the system. Figure 1 The comparison is a detailed description of the sports training system based on the visual AI model disclosed in this application.
[0067] Before describing the embodiments of the present application, the visual AI model shown in the present application is a two-dimensional human posture estimation model, which can detect the positions of key points of the human body (such as nose, eyes, shoulders or various joints, etc.) from RGB images or videos. Its basic method is to first detect the human body boundary box, and then estimate the posture of the human body in each boundary box separately, so as to obtain relevant data and correctness of the user's posture, etc.
[0068] Specifically, the visual AI model proposed in this application may include the following steps when performing sports training: 1. Posture detection: First, use human posture detection technology (see relevant prior art) to process the video frames of the user's movement process, detect the user's key point coordinates (key points such as joint positions), and process the movement process in real time or offline. The recorded video. 2. Posture evaluation: According to the detected key point information, evaluate whether the user's posture is correct. This requires pre-defining a reference template for a standard action, such as the angle range of each joint, the relative position of the limbs, etc., and comparing the user's actual posture with the standard template to determine whether the posture is correct. 3. Feedback prompts: According to the results of the posture evaluation, give corresponding feedback and guidance, and the user can be prompted on the interface in the form of text, images, voice, etc. to indicate the parts that need to be adjusted and the correct action essentials. 4. Action recognition: In addition to posture analysis for a single frame, a continuous action sequence can also be identified. By analyzing the motion trajectory, angle and time changes of the key points, it can be determined which specific action the user is doing, such as push-ups, squats, etc. 5. Action counting: Based on the identification of the action type, the number of times each action is completed can be further counted. This is generally achieved by detecting the periodic changes of key points, such as the rise and fall of the elbow angle during push-ups; action counting can help users quantify exercise intensity and set reasonable exercise goals. 6. Historical tracking: The posture evaluation results and action statistics of each exercise are saved to form a time series; users can review their historical performance, understand the mastery of technology and progress trends, and adjust training plans based on these data. 7. Human-computer interaction: In addition to post-analysis, posture tracking technology can also be used to design various interactive training systems; for example, a virtual picture is displayed on the screen to guide users to make movements or the user's movements are mapped to the game character in real time, thereby increasing the fun and immersion of the exercise.
[0069] First, the sports training system based on the visual AI model includes a mobile terminal and a server terminal, wherein the mobile terminal is a client terminal, which can be a mobile device such as a mobile phone or a PAD, and the server terminal is a server terminal, which can be one or more combinations of a WEB server, a database server, and a cloud server, etc., which will not be elaborated in detail here. The client of this application collects and processes data for the user's sports training based on the visual AI model. The mobile terminal disclosed in this application collects user information based on the visual AI model. The user information can be the ID card number, name, gender, and other information when the user logs in to the system through the mobile terminal, or it can be the face information collected by the camera of the mobile terminal or the fingerprint information collected by the fingerprint module. Of course, it can also be collected by first collecting the ID card number, name, gender, and other information logged in by the mobile terminal, and then the face information collected by the mobile terminal camera, and after comparing the two, it is determined whether the two are unified, and then the next step of related operations is performed. That is, the mobile terminal obtains the user's training data information according to the user information and collects the dynamic training data of the user's sports training process in real time. The mobile terminal of this application collects corresponding information based on the visual AI model.
[0070] Of course, in the specific implementation process, the mobile terminal can register, log in, modify and retrieve accounts through the user's ID number, avatar, etc., and each person can set up multiple accounts, and multiple accounts can also be associated with each other, and personal information can also be changed accordingly, etc. At the same time, in the specific implementation process, the mobile terminal can also have calendar punch-in, plan creation / management, data upload, payment and other related functions. The relevant functions can refer to the relevant description of the existing technology, and will not be elaborated in detail here. Even in the implementation process of this application, since the camera is required to capture the user's movements (or postures) during sports training in real time, in order to judge the captured posture (such as joint movement angle, etc.), it is necessary to animate the user's sports training movements to form corresponding animations and thus obtain corresponding data, so that the mobile terminal or server side can perform corresponding data processing. At the same time, during the server-side processing, the actions in the preset database can be compared with the data obtained from the captured posture to judge the effectiveness or accuracy of the user's actions, and the current user's actions can be played by voice, etc. Of course, the above is only an exemplary explanation and will not be elaborated here. The following is a relevant description of the embodiments of the present application in combination with a sports training system based on a visual AI model, a sports training device based on a visual AI model, and a sports training method based on a visual AI model.
[0071] The server side is used to draw the user's actual motion curve based on the collected joint angles and conclusion data. The server side is also used to calculate the actual motion progress information of the user after exercise training based on the actual motion curve, the user's training times, the current planned motion progress information and the current motion progress information. The server side is also used to correct the above-mentioned current planned motion progress information based on the actual motion progress information and the number of exercise trainings to obtain the corrected planned motion progress information to guide the user in exercise training.
[0072] In the specific implementation process, the motion training method based on the visual AI model is also applied to a motion training device based on the visual AI model, such as Figure 2 As shown, the sports training device may include an identity acquisition module 11, an information acquisition module 12, a data acquisition module 13, a curve drawing module 14, an information calculation module 15 and an information correction module 16, wherein the identity acquisition module 11 is used to collect user information; the information acquisition module 12 is used to obtain the user's training data information based on the user information; the data acquisition module 13 is used to collect dynamic training data of the user during the sports training process in real time; the curve drawing module 14 is used to draw the user's actual motion curve based on the collected joint angles and conclusion data; the information calculation module 15 is used to calculate the actual motion progress information of the user after the sports training based on the actual motion curve, the user's training times, the current planned motion progress information and the current motion progress information; the information correction module 16 is used to correct the above-mentioned current planned motion progress information based on the actual motion progress information and the sports training times, and obtain the corrected planned motion progress information to guide the user to perform sports training.
[0073] In combination with the above disclosed sports training device based on visual AI model and sports training system based on visual AI model, this application is combined with the attached Figure 3-10 An implementation of a motion training method based on a visual AI model is also disclosed, and details can be found in the following description.
[0074] like Figure 3 As shown, the motion training method based on the visual AI model disclosed in this application includes:
[0075] In step S100, user information of the user to be trained is collected through a mobile terminal.
[0076] During the specific implementation process, the user information can be the ID number, name, gender, etc. of the user when logging into the system through the mobile terminal, or it can be the face information collected by the mobile terminal camera or the fingerprint information collected by the fingerprint module. Of course, during the specific implementation process, since the user needs to log in to the mobile APP system first, the mobile terminal can at least obtain the user's fingerprint or face information after logging into the system, and then obtain the user's ID number, name, gender and other information corresponding to the feature based on the fingerprint and face information. The details are not elaborated here.
[0077] In step S200, the user's training data information is obtained according to the user information.
[0078] During the specific implementation process, since the user has a corresponding case record in the hospital or has recorded the corresponding exercise training information in the system before using the system, after obtaining the user information in step S100, some training information corresponding to the user stored in the database can be retrieved through the user information. The training data information includes the user's exercise training times (i.e., the number of exercise trainings that have been performed, of course, this number can also be 0), the current planned exercise progress information (i.e., the planned exercise progress information before this exercise training, which may include the planned exercise training times, the planned exercise training time, and the degree of joint recovery of the planned exercise, etc.) and the current exercise progress information (i.e., the exercise progress before this exercise training).
[0079] In step S300, dynamic training data of the user during the exercise training process is collected in real time.
[0080] During the specific implementation process, after the user logs in to the client, he can perform corresponding movements according to the corresponding planned movement information given by the visual AI model, such as the movement of the user's disabled joints, so as to collect the user's movements in the movement training process based on the visual AI model, so that dynamic training data can be collected. Specifically, the dynamic training data includes the joint angles during the user's movement training process and the conclusion data given by the visual AI model for the joint angles. Among them, the conclusion data is a corresponding conclusion of whether the joint angles of the user's active joints during the movement meet the requirements through the visual AI model, thereby judging whether the data is valid.
[0081] Of course, during the specific implementation process, the conclusion data can also be the degree of completion of the joint angle (such as completion percentage). For example, the joint angle required for sports training is 100 degrees, and the actual completed joint angle is 99 degrees, then the conclusion data is 99%. The above is only an exemplary representation and is not a technical solution that must be adopted in this application. It is only shown here as an example and will not be elaborated in detail.
[0082] In step S400, the actual motion curve of the user is drawn according to the collected joint angles and conclusion data.
[0083] For specific implementation, please refer to the attached Figure 4 , the attached Figure 4 Detailed steps of step S400 are shown.
[0084] In step S401, function recovery index data related to user information is obtained.
[0085] In the specific implementation process, the function recovery index data refers to an index data that the user can achieve after exercise through exercise training, which may include a predetermined value that the user's joint angle can reach after exercise training, etc. Among them, the function recovery index data is the relevant data before this exercise training.
[0086] In step S402, an actual motion curve including a P value (with a statistical significance of less than 0.05) with the horizontal axis being time and the vertical axis being motion state is obtained by statistically calculating the joint angle, conclusion data and functional recovery index data according to a statistical algorithm.
[0087] During the specific implementation process, the time in the coordinate system is in days, weeks or months; the functional recovery index data at least includes the recovery angle of the disabled joint, muscle strength, mobility, and imaging evaluation results. Among them, the mobility mainly refers to the type and intensity of normal joint activities, and the imaging evaluation results mainly refer to the relevant results of X-rays or MRI, showing the situation after the fracture is healed. The conclusion data is the conclusion of whether the camera captures the user's sports training movements or postures in real time to determine whether the joint angle is correct.
[0088] During the specific implementation process, there are three commonly used methods for the statistical algorithm of this application, 1) difference analysis (mainly studying the differences between different data, including single sample mean comparison, two sample mean comparison, multiple sample mean comparison, and categorical data comparison), 2) correlation study (mainly studying the correlation between data, including using the Pearson correlation coefficient for analysis when the normal distribution is satisfied, using the spearman correlation coefficient for analysis when the normal distribution is not satisfied, and using the spearman correlation coefficient for analysis when there are multiple classified ordered data), 3) influencing factor research (mainly studying the influencing factors of the dependent variable, according to the different types of dependent variables, it can be divided into quantitative research - using linear regression method for research, categorical research using logistic regression method for research, and survival data research using cox research method). After we obtain the relevant parameters of the user, we can use appropriate statistical research methods to obtain joint angles, conclusion data, and functional recovery index data based on the relevant parameters to calculate the corresponding actual motion curve containing the P value. The specific application of the above statistical algorithm is only a way to illustrate, and is not limited to this way. In addition, this application only applies the relevant statistical algorithm in the prior art to calculate the corresponding horizontal coordinate and total coordinate curves, which will not be elaborated in detail here.
[0089] At the same time, in the specific implementation process, this statistical algorithm is only an exemplary statistical strategy. First, it can include the following prediction models: logistic prediction model and Cox regression to construct a prediction model, and then the above model can use the Lasso algorithm to actually carry out the prediction model; second, conduct real-world research, including the key points of retrospective study design, the bias of observational studies and its control, common statistical methods of observational studies, principles and methods of independent variable screening methods, propensity score methods, propensity score weighted matching, trend research analysis methods, and observational research paper writing methods, which are not elaborated in detail here.
[0090] Of course, during the specific implementation process, in addition to referring to the joint angles, conclusion data after capturing postures and functional recovery indicators, the statistical algorithm can also use the user's daily movements, movement angles, effective times, frequency (for example, 9 times a day, 3 times in the morning, 3 times at noon, 3 times in the evening, 10 sets of movements a day, what angle should be achieved in each set of movements, etc.), and other data to statistically calculate the corresponding actual motion curve. The details will not be elaborated here.
[0091] In step S500, the actual exercise progress information of the user after exercise training is calculated according to the actual exercise curve, the number of training times of the user, the current planned exercise progress information and the current exercise progress information.
[0092] For specific implementation, please refer to the attached Figure 5, after getting the actual motion curve, you can perform the following steps:
[0093] In step S501, the functional recovery index data of the user after the current exercise training is obtained according to the actual exercise curve.
[0094] During the specific implementation process, the functional recovery index data at least includes the recovery angle of the disabled joint, muscle strength, mobility, and imaging assessment results. The functional recovery index data described in step S501 is the index data obtained after this sports training, which is different from the functional recovery index data described in step S401.
[0095] In step S502, the functional recovery index data of the user after the current exercise training is compared with the dynamic training data in the current exercise progress information.
[0096] During the specific implementation process, the dynamic training data in the current exercise progress information generally refers to the dynamic training data that the disabled joint should be able to achieve a certain degree of recovery if the current exercise training is carried out. Therefore, it is necessary to compare the functional recovery index data after this exercise training with the dynamic training data in the current exercise progress information to see whether there is a corresponding difference between the two or how much the gap is.
[0097] In step S503, the actual exercise progress information of the user after the exercise training is obtained according to the comparison result combined with the number of training times of the user and the current planned exercise progress information.
[0098] During the specific implementation process, step S502 obtains the difference or distinction between the functional recovery index data after this exercise training and the dynamic training data in the current exercise progress information. Then, combining the comparison result with the number of exercise training times the user has already performed and the current planned exercise progress information (such as the dynamic training data of the planned exercise, etc.), the actual recovery progress information of the user after the exercise training can be obtained. For example, the user's exercise recovery progress is compared with the planned exercise progress. If it is below 100%, it means that the degree of recovery from the exercise training is low. If it is higher than 100%, it means that the degree of recovery from the exercise training is high or far exceeds expectations. The above is only an exemplary description and will not be elaborated here.
[0099] In step S600, the current planned exercise progress information is corrected according to the actual exercise progress information and the number of exercise trainings to obtain the corrected planned exercise progress information to guide the user to perform exercise training.
[0100] In the specific implementation process, by correcting the current planned exercise progress information, it is possible to effectively guide the user's next exercise training process. For example, according to step S503, if the exercise training recovery level is low, then the exercise training time and / or exercise training intensity may be increased in the next exercise training plan, otherwise the exercise training time or exercise training intensity may be reduced according to the original plan. The details are not elaborated again. Of course, you can also refer to the attached Figure 6 The detailed steps of step S600 are described.
[0101] In step S601, the planned exercise training times, planned exercise training time and planned exercise training index data in the current planned exercise progress information are obtained.
[0102] In the specific implementation process, after completing step S503, it is necessary to know the planned number of exercise training, planned exercise training time and planned exercise training index data in the current planned exercise progress information. Here, the applicant needs to explain that the current planned exercise progress information is the exercise progress information before this exercise training (i.e. after the last exercise training), and the actual exercise progress information is the exercise progress information after this exercise training. By understanding the planned number of exercise training, planned exercise training time and planned exercise training index data in the last planned exercise progress information, it is possible to make corresponding adjustments to the index data after this exercise training, so that the user's subsequent exercise training can be guided after each exercise training, so that the user's exercise training movements and time can be adjusted in time, so as to achieve the exercise effect as soon as possible.
[0103] In step S602, the activity angle of the disabled joint after the actual exercise training is obtained according to the actual exercise progress information.
[0104] During the specific implementation process, the activity angle of the disabled joint refers to the movement standard that the disabled joint can achieve after this sports training or the movement goal achieved by the disabled joint. The activity angle of the disabled joint can usually be stored in a database, so the activity angle can be obtained through actual sports progress information.
[0105] In step S603, the planned exercise training times and the planned exercise training time are corrected according to the activity angle and the exercise training times, and new planned exercise training time, planned exercise training times and planned exercise training indicator data are obtained as corrected planned exercise progress information.
[0106] During the specific implementation process, when the activity angle of the user's disabled joint and the number of exercise training sessions that have been performed are obtained, the activity angle and the number of exercise training sessions can be compared with the dynamic training data and the planned exercise training times in the previous planned exercise progress information, and then the planned exercise training times and the planned exercise training time can be corrected to obtain new planned exercise training time, planned exercise training times and planned exercise training indicator data. For example: the dynamic training data in the last planned exercise progress information is an activity angle of 10 degrees, a planned number of exercise training times of 28 times, and a planned exercise training time of 14 days. According to the actual exercise training information, the activity angle of the user's disabled joint is 8 degrees and the number of exercise training times that have been performed is 16 times. The correction method can be that the activity angle / exercise training times (8 / 16) in the actual exercise training information is 0.5 degrees, that is, the activity angle of the disabled joint is 0.5 degrees for each exercise training in the actual exercise training. The planned exercise training times in the new planned exercise training information can be the activity angle in the last planned exercise progress information minus the activity angle in the actual exercise training information (10-8), which is 2 degrees. The planned exercise training times in the new planned exercise training information are (2 / 0.5) 4 times. If exercise training is performed twice a day, the new planned exercise time is (4 / 2) 2 days. Of course, the above is only an exemplary description, and does not mean that this training method is also followed during the implementation of this application. Moreover, the relevant information such as the activity angles of 10 degrees and 8 degrees described in the above embodiment does not represent a certain user, but is an exemplary description. In the actual process, it is still necessary to set and explain according to the actual situation, which will not be elaborated here.
[0107] Of course, in the specific implementation process, you can also refer to the attached Figure 7 A flow chart of a motion training method based on a visual AI model disclosed in . The steps include:
[0108] In step S6031, the number of exercise training times and the exercise training time corresponding to the activity angle and the number of exercise training times are searched in a preset rule database.
[0109] During the specific implementation process, the preset rule database can be a database established according to the joints and corresponding movement movements. Of course, it can also be a database related to the activity angle and the number of movement training added on the basis of the above database, which will not be elaborated in detail here.
[0110] In step S6032, the number of exercise training sessions and the exercise training time are compared with the planned number of exercise training sessions and the planned exercise training time, and the difference between the two is calculated respectively.
[0111] During the specific implementation process, if the normal exercise training is performed according to the exercise training plan, the number of exercise training times remaining after each exercise training is usually equal to the exercise training time. However, due to the individual differences of each user and the failure to perform exercise according to the plan or the action performed during the exercise training process not meeting the regulations, there will be differences between the two. Therefore, the number of exercise training times and the exercise training time obtained by step S6031 in this application are compared with the planned number of exercise training times and the planned exercise training time, and the difference between the two can be obtained. If the difference is a positive value, it means that the user does not perform exercise according to the plan or the exercise action is not compliant, etc., then it is necessary to increase the planned number of exercise training times and the planned exercise training time, otherwise it may be necessary to reduce the corresponding number and time.
[0112] In step S6033, the planned exercise training times and planned exercise training times are added to half of the difference between the two and the integer is taken to obtain the new planned exercise training time and planned exercise training times.
[0113] During the specific implementation process, please refer to the corresponding instructions of step S6032. It is necessary to take integers to calculate the new planned exercise training time and planned exercise training times. Since the difference between the two may be a negative number, the new time and times may increase or decrease, which will not be elaborated in detail here.
[0114] The modification scheme for the planned exercise progress information disclosed in the embodiment of the present application is only an exemplary description given in the present application, and can be specifically combined with the above-mentioned appendix. Figure 6 The relevant schemes disclosed are only described and will not be elaborated in detail here.
[0115] In addition, see Appendix Figure 8 FIG. 1 is a flow chart of a motion training method based on a visual AI model. This is mainly a further detailed description of step S300. Figure 8 It can be seen that step S300 disclosed in this application also includes the following steps:
[0116] In step S301, the joint angle of the user's exercise training and the time for maintaining the joint angle are obtained in real time.
[0117] During the specific implementation process, when the user performs sports training, the training goal of this sports training (such as the joint angle to be achieved in this training and the time to maintain the joint angle, etc.) can usually be sent out through the mobile terminal. The user can stretch the joints according to the training goal, and the stretched joints can obtain the corresponding joint angle information after real-time collection and calculation by the mobile terminal. At the same time, the holding time of the joint angle can also be obtained, etc., which will not be elaborated here.
[0118] In step S302, data association is performed based on a combination of a generative model and recognition capture.
[0119] In the specific implementation process, the generative model, namely generative artificial intelligence (Generative AI), is an artificial intelligence technology that can autonomously generate text, images, audio and other content. For related technologies, please refer to the prior art, which will not be elaborated in detail here. Recognition capture is to capture the posture of the user in the process of sports training in real time through the camera. In this application, we can first obtain the video and angle information of the user through digital guidance of the sports group outside the hospital. At the same time, in order to better integrate with the generative model, we need to communicate with Yuan on the collection of muscle electrical signals. Through the collection of muscle electrical signals, the visual AI model and the generative model can be integrated with each other to form a dynamic action based on generative (that is, the AI training action used by the user for sports training). The above is only an exemplary explanation. Please refer to the description before and after the specification and the relevant content of the prior art for details. It will not be elaborated here.
[0120] In step S303, the joint angles of the user's exercise training and the time the joint angles are maintained are compared according to the data in the preset rule database, and the joint angles of the user's exercise training and the time the joint angles are maintained are determined and analyzed.
[0121] During the specific implementation process, since the generative model does not provide a specific rule judgment for a certain angle of joint activity, this requires us to optimize and give the model relevant rules, so that the visual AI model can compare and analyze the actual captured values according to the theoretical values we provide, so as to judge the correctness or error of the actual values. Specifically, a corresponding rule database can be established in advance for the theoretical value, and by comparing the actual dynamic training data with the theoretical value data in the preset rule database, it can be determined whether the joint activity angle corresponding to the actual dynamic training data is valid or correct. Of course, the above is only an exemplary description. During the specific implementation process, each joint action can also correspond to one or more theoretical values in the preset rule database (different age groups, such as 1-10 years old, 11-14 years old, 15-18 years old, 19-22 years old, 23-30 years old, 30-40 years old, etc., all have corresponding theoretical values. The above is only an exemplary description and does not represent the scope of protection of this application).
[0122] During the specific implementation process, the pre-established corresponding rule database may also include maintaining each joint angle and the corresponding time required to maintain each joint angle, so that the joint angle collected in step S301 and the time to maintain the joint angle can be compared with the data in the preset rule database to obtain the degree of completion of the joint angle that can be maintained for the corresponding time or whether the joint activity angle is effective or correct, which can be used as the conclusion data given for the joint angle by the visual AI model, and will not be elaborated here.
[0123] In step S304, the joint angles that conform to the data in the preset rule database and the conclusion data given by the visual AI model for the joint angles are used as dynamic training data to be collected during the user's motion training process.
[0124] During the specific implementation process, when the corresponding conclusion is obtained in step S303, the actual dynamic training data for which the judgment result is valid or correct can be determined to be valid data, that is, data that can affect sports training or be helpful for sports training, thereby helping users to perform sports training in a timely and effective manner during subsequent user sports training.
[0125] At the same time, in the specific implementation process, please refer to the attached Fig. 9 As shown, the present application also discloses a motion training method based on a visual AI model, comprising the following steps:
[0126] In step S700, the training data of each exercise training of the user is obtained through the server, and the training data includes the angle, position, number of times and qualified number of disabled joint exercise training.
[0127] During the specific implementation process, when the user is performing sports training, the camera will collect the user's relevant movements, such as the position of the joints (such as collecting the position of the active joints - upper limbs, lower limbs or elbow joints, etc.), the angle of joint movement, the number of joint movements and the relatively qualified number of times. The specific details will not be elaborated here.
[0128] In step S800, the training data is classified.
[0129] In the specific implementation process, the application can establish a database according to the joints and the corresponding sports movements, that is, the angle, position, number of times and qualified number of each training of a joint during sports training are collected and stored in the database. Of course, it can also be classified and processed accordingly through an EXCEL table, but the EXCEL table is only a form of recording. The core is to establish a database such as video picture angles when the user performs corresponding actions according to the sports movements matched by different joints.
[0130] In step S900, the classified training data, actual motion curve and actual motion progress information are trained to obtain key data affecting the motion training, and a corresponding preset rule database is established according to the key data.
[0131] In the specific implementation process, the classified training data will be used for research similar to randomized controlled trials (RCTs) later, such as a random group and a control group, changing a parameter in the database to see if it affects the effect. Through relevant experiments and data processing, key data that affects sports training can be obtained, so that the corresponding data can be used to guide users' sports training.
[0132] For specific implementation process, please refer to the attached Fig.10 The detailed flow chart of step S900 disclosed in this application is shown in FIG. Fig.10 The step S900 further includes the following steps:
[0133] In step S901, the disabled joints of the user are detected to obtain the disability information of the user's joints.
[0134] During the specific implementation process, the detection of the user's disabled joints can be carried out by experts through the relevant results of the user's X-ray or MRI, thereby determining the relevant information of the user's disabled joints, such as the disabled joint movement angle and the position of the disabled joint, etc., which will not be elaborated here, and please refer to the description of the above implementation scheme.
[0135] In step S902, a planned movement value is obtained by comparing the disability information with a preset rule database including at least a movement knowledge graph or standard movement movements.
[0136] During the specific implementation process, the preset rule database at least includes a motion knowledge graph and standard motion movements, wherein the motion knowledge graph and standard motion movements can generally be collected through the database from existing books and medical knowledge related information on motion knowledge corresponding to different disabled joints: for example, standard motion movements, motion cycles, motion angles required for sports training, and corresponding graph movements, etc. However, in the actual operation process, the database can also continuously optimize and adjust the motion knowledge graph and standard motion movements through the continuously collected big data content of the user's actual motion. This application will not elaborate on this, and please refer to the prior art, etc.
[0137] Therefore, after obtaining the disability information, the movement angle of the disabled joint and the position of the disabled joint can be known. Since the joints of users of different age groups have a corresponding range of motion under normal conditions, the movement angle of the joint under normal conditions and the position of the disabled joint can be combined with the user information to obtain the movement angle of the joint that needs to be moved at the position of the disabled joint. For example: Under normal circumstances, the joint of a user before an elbow fracture can move 180 degrees, but the joint of the user can only move to 150 degrees after the current disability. Then the difference in the required movement is 180-150, which is equal to 30 degrees, that is, the planned movement value is the user's elbow joint planned movement of 30 degrees. Of course, the above is only an exemplary description, and the specific needs are determined according to the user's age, gender, joint position, etc., which will not be elaborated in detail here.
[0138] In step S903, the required exercise cycle and training content are obtained according to the planned exercise values, and the angle of the joint exercise training required by the user every day is obtained, and it is used as key data affecting exercise training to guide the user to perform exercise training.
[0139] During the specific implementation process, if the planned exercise value obtained in step S902 is 30 degrees, then the required exercise cycle and training content can be calculated based on the degree of exercise that can be recovered every day during the actual exercise process. For example, if 1 degree of exercise can be achieved every day, then the exercise cycle is 30 / 1=30 days, and the training content is the exercise movements required to achieve 1 degree of exercise, such as each person doing 10 exercises at 2 angles every day, etc. It will not be elaborated here, that is, the training content is the exercise movements (such as activity angles, etc.) that users need to achieve their exercise goals every day.
[0140] In the specific implementation process, the existing planned exercise progress is based on the content required by the exercise course. This is only a theoretical value, and there is no actual user indicator as a reference, that is, there is no positive correlation between the theoretical value and the actual value. The theoretical value must be further optimized and improved through a large amount of real user data. For example, the theoretical value is based on each person doing 10 2-angle exercises every day, but after a large amount of actual user data is collected, after statistical analysis (such as RCT research or R language statistical conclusions), it may be concluded that each person only needs to do 1-2 4-angle exercises every day to improve joint function training, thereby optimizing the theoretical value. Now there is only consensus but no guidelines in the domestic medical community for exercise. What is missing from consensus to clinical guidelines is similar medical evidence. Through this set of mobile-based visual technology, we can have more similar follow-up data, allowing hospitals to realize relevant data verification and mining outside the hospital.
[0141] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A motion training method based on a visual AI model, characterized in that: include: Collect user information of users to be trained through mobile terminals; Acquire the user's training data information according to the user information, wherein the training data information includes the user's exercise training times, current planned exercise progress information and current exercise progress information; Real-time collection of dynamic training data during the user's sports training process, wherein the dynamic training data includes the joint angles during the user's sports training process and conclusion data of the joint angles given by the visual AI model; Draw the actual motion curve of the user according to the collected joint angle and conclusion data; wherein, this step includes: obtaining functional recovery index data related to the user information; obtaining an actual motion curve with the horizontal axis being time and the vertical axis being motion state, including a P value, by statistically calculating the joint angle, conclusion data and functional recovery index data according to a statistical algorithm; wherein, the time is in days, weeks or months; the functional recovery index data at least includes the recovery angle of the disabled joint, muscle strength, activity ability, and imaging evaluation results; Calculate the actual exercise progress information of the user after the exercise training according to the actual exercise curve, the number of training times of the user, the current planned exercise progress information and the current exercise progress information; The current planned exercise progress information is corrected according to the actual exercise progress information and the number of exercise trainings to obtain the corrected planned exercise progress information to guide the user to perform exercise training; wherein the correction step further includes: Obtain the planned exercise training times, planned exercise training time and planned exercise training indicator data in the current planned exercise progress information; Acquire the activity angle of the disabled joint after actual sports training according to the actual sports progress information; The planned exercise training times and the planned exercise training time are modified according to the activity angle and the exercise training times, and new planned exercise training time, planned exercise training times and planned exercise training indicator data are obtained as modified planned exercise progress information; wherein the modified planned exercise training times and the planned exercise training time are further modified according to the activity angle and the exercise training times: Searching the preset rule database for the number of exercise training and the exercise training time corresponding to the activity angle and the number of exercise training respectively; Compare the exercise training times and exercise training time with the planned exercise training times and planned exercise training time, and calculate the difference between the two respectively; The planned exercise training times and the planned exercise training time are added to half of the difference between the two and the integer is taken to obtain the new planned exercise training time and the planned exercise training times; The establishment of the preset rule database includes the following steps: Acquire the training data of each exercise training of the user through the server, wherein the training data includes the angle, position, number of times and the number of qualified times of the exercise training of the disabled joint; Classify the training data; Training the classified training data, actual motion curves, and actual motion progress information to obtain key data affecting the motion training, and establishing a corresponding preset rule database based on the key data; The classified training data, the actual motion curve and the actual motion progress information are trained, including: Detecting disabled joints of the user to obtain disability information of the user's joints; Obtaining a planned movement value by comparing the disability information with a preset rule database including at least a movement knowledge graph or standard movement actions; According to the planned exercise values, the required exercise cycle and training content are obtained, and the angle of the joint exercise training required by the user every day is obtained, and it is used as the key data affecting the exercise training, so as to guide the user to perform exercise training; Among them, the preset rule database at least includes a motion knowledge graph and standard motion movements. The motion knowledge graph and standard motion movements collect relevant information of motion knowledge corresponding to different disabled joints in existing books and industry knowledge through the database. The relevant information includes standard motion movements, motion cycles, motion angles required for sports training, and corresponding graph movements. The preset rule database continuously optimizes and adjusts the motion knowledge graph and standard motion movements through the continuously collected big data content of users' actual movements.
2. The motion training method based on the visual AI model according to claim 1, characterized in that: According to the actual exercise curve, the number of training times of the user, the current planned exercise progress information and the current exercise progress information, the actual exercise progress information of the user after the exercise training is obtained, including: Acquiring functional recovery index data of the user after the current exercise training according to the actual exercise curve; Compare the user's functional recovery index data after the current exercise training with the dynamic training data in the current exercise progress information; According to the comparison result, combined with the user's training times and the current planned exercise progress information, the actual exercise progress information of the user after exercise training is obtained.
3. The motion training method based on the visual AI model according to claim 1, characterized in that: Real-time collection of dynamic training data during user exercise training, including: Obtain the joint angle of the user's exercise training and the time to maintain the joint angle in real time; Data association is performed based on a combination of generative models and recognition capture; Comparing the joint angles of the user's exercise training and the time the joint angles are maintained according to the data in the preset rule database, and making a judgment and analysis on the joint angles of the user's exercise training and the time the joint angles are maintained; The joint angles that conform to the data in the preset rule database and the conclusion data given by the visual AI model on the joint angles are used as the dynamic training data to be collected during the user's motion training process.
4. A sports training device based on a visual AI model, using the sports training method based on a visual AI model as claimed in claim 1, characterized in that: include: An identity collection module, used to collect user information of the user to be trained; An information acquisition module, used to acquire the user's training data information according to the user information; The data acquisition module is used to collect dynamic training data of the user during the sports training process in real time; The curve drawing module is used to draw the user's actual motion curve based on the collected joint angles and conclusion data; An information calculation module, used to calculate the actual exercise progress information of the user after exercise training according to the actual exercise curve, the number of training times of the user, the current planned exercise progress information and the current exercise progress information; The information correction module is used to correct the above-mentioned current planned exercise progress information according to the actual exercise progress information and the number of exercise trainings, and obtain the corrected planned exercise progress information to guide the user to perform exercise training.
5. A sports training system based on a visual AI model, using the sports training method based on a visual AI model as claimed in claim 1, characterized in that: It includes mobile terminal and server terminal, among which, The mobile terminal collects user information of the user to be trained based on the visual AI model, and the mobile terminal is also used to obtain the user's training data information according to the user information and to collect dynamic training data of the user during the training process in real time; The server is used to draw the user's actual motion curve based on the collected joint angles and conclusion data. The server is also used to calculate the actual motion progress information of the user after exercise training based on the actual motion curve, the user's training times, the current planned motion progress information and the current motion progress information. The server is also used to correct the above-mentioned current planned motion progress information based on the actual motion progress information and the number of exercise trainings to obtain the corrected planned motion progress information to guide the user in exercise training.
Citation Information
Patent Citations
Rehabilitation training plan generation and rehabilitation training system
CN115841863A
Military training method and system based on human body posture recognition
CN115937894A
Lower limb rehabilitation training equipment, lower limb rehabilitation training method and storage medium
CN117711568A
Fitness optimization training method and system based on virtual reality technology
CN117766098A
Method and system for generating individual training sequences in a simulation environment, especially for sports and rehabilitation purposes.
DE102019120655A1
