Fitness information pushing method, model training method and related equipment

Abstract

The invention provides a fitness information pushing method, a model training method and related equipment, which are applied to the technical field of electrical digital processing, and the method comprises the following steps: obtaining a current body type modeling and a fitness plan of a user; on the basis of the fitness plan, biomechanical parameters are determined, and the biomechanical parameters are used for indicating the relation between a calorie notch caused by the fitness plan to the human body of the user and the lipolysis rate; the current body shape modeling and the biomechanical parameters are input into a pre-trained target model, predicted body shape modeling of the user is obtained, and the target model is used for analyzing the influence of the biomechanical parameters on the current body shape modeling so as to output the predicted body shape modeling; and pushing the predicted body type modeling to the user. In the method, compared with the prior art, the obtained prediction body type modeling is more accurate, that is, the accuracy of the pushed information can be improved by the method provided by the embodiment of the invention.

Description

Technical Field

[0001] This application relates to the field of electronic digital processing technology, and in particular to a fitness information push method, a model training method, and related equipment. Background Technology

[0002] Existing fitness information pushes come in various forms, such as using artificial intelligence (AI) to predict body shape and push body shape information; or using AI to create fitness plans and push fitness plan information.

[0003] In the field of AI body shape prediction, existing technologies primarily rely on basic user-input data (such as height, weight, and age) to generate static predictions through machine learning algorithms. For example, some AI fitness assistants analyze parameters like body fat percentage and muscle mass to generate a three-dimensional (3D) model of the user's current body shape. However, current technologies cannot simulate the body shape evolution during fat loss or muscle gain, thus failing to deliver accurate body shape information. Therefore, existing fitness information delivery methods suffer from inaccurate information delivery. Summary of the Invention

[0004] This application provides a fitness information push method, a model training method, and related equipment, which can solve the problem that existing fitness information push methods are not accurate enough.

[0005] In a first aspect, embodiments of this application provide a method for pushing fitness information, the method comprising: Obtain the user's current body shape model and fitness plan; Based on the fitness plan, biomechanical parameters are determined, which are used to indicate the relationship between the calorie deficit and fat breakdown rate caused by the fitness plan to the user's body. The current body shape model and the biomechanical parameters are input into a pre-trained target model to obtain the user's predicted body shape model. The target model is used to analyze the influence of the biomechanical parameters on the current body shape model in order to output the predicted body shape model. The predicted body shape model is pushed to the user.

[0006] Optionally, after obtaining the user's current body shape model and fitness plan, the method further includes: Upon receiving an adjustment operation from the user regarding the current body shape model, obtain the desired body shape model obtained based on the adjustment operation. The current body shape model, the fitness plan, and the desired body shape model are input into the target model to obtain the modified fitness plan. The target model is used to analyze the current body shape model and the desired body shape model to adjust the fitness plan and obtain the modified fitness plan. The modified fitness plan is pushed to the user.

[0007] Optionally, the target model is further used for: Extract the time-series features of the fitness plan; Based on the difference between the current body shape model and the desired body shape model, the deformation characteristics are determined; The time series features and the deformation features are encoded into a unified dimension tensor to obtain the first encoded data corresponding to the time series features and the second encoded data corresponding to the deformation features; Based on preset physiological tolerance constraint parameters, the first encoded data, and the second encoded data, the fitness plan is adjusted to obtain the modified fitness plan. The physiological tolerance constraint parameters are used to ensure that the user's body shape changes conform to physiological laws.

[0008] Optionally, determining biomechanical parameters based on the fitness plan includes: Convert the fitness plan into exercise parameters; Obtain a first function from a pre-defined biomechanical simulation library. This first function is used to indicate the force generated by human muscle groups at different lengths and contraction speeds. The biomechanical parameters are determined based on the motion parameters and the first function.

[0009] Optionally, obtaining the user's current body shape model includes: Acquire first data and second data. The first data is three-dimensional point cloud data generated after three-dimensional scanning of the user's human body. The second data is a human body volume density model obtained after performing bioelectrical impedance analysis and human body displacement air volume analysis on the user's human body. The first data is adjusted based on the second data to obtain the current body shape model.

[0010] Secondly, embodiments of this application also provide a model training method, the method comprising: Obtain the preset model and training dataset; The preset model is trained based on the training dataset to obtain the target model as described in the first aspect. The target model is used to analyze the influence of the input biomechanical parameters on the current body shape modeling to output a predicted body shape modeling.

[0011] Optionally, the preset model includes a first sub-model, and the training dataset includes a first subset, which includes historical current body shape modeling, historical biomechanical parameters, and real body shape modeling corresponding to the historical current body shape modeling and the historical biomechanical parameters. The step of training the preset model based on the training dataset to obtain the target model includes: Using the historical current body shape model and the historical biomechanical parameters as input data, and the real body shape model as the label, the first sub-model is trained to obtain a body shape dynamic prediction sub-model. The first loss function during the training of the first sub-model is determined based on the real body shape model and the predicted body shape model output by the first sub-model. The body shape dynamic prediction sub-model is used to analyze the influence of the input biomechanical parameters on the input current body shape model in order to output a predicted body shape model.

[0012] Optionally, the preset model includes a second sub-model, and the training dataset includes a second subset, which includes historical current body shape modeling, historical fitness plans, historical expected body shape modeling, and real fitness plans corresponding to the historical current body shape modeling, the historical fitness plans, and the historical expected body shape modeling. The step of training the preset model based on the training dataset to obtain the target model includes: The historical current body shape model, the historical fitness plan, and the historical desired body shape model are used as input data, and the actual fitness plan is used as a label to train the second sub-model to obtain a fitness plan adjustment sub-model. The fitness plan adjustment sub-model is used to analyze the input current body shape model and the input desired body shape model in order to adjust the input fitness plan and obtain a modified fitness plan. The second loss function during the training of the second sub-model is determined based on the real fitness plan and the modified fitness plan output by the second sub-model.

[0013] Thirdly, embodiments of this application also provide a fitness information push device, the device comprising: The first acquisition module is used to acquire the user's current body shape model and fitness plan; The first determining module is used to determine biomechanical parameters based on the fitness plan, wherein the biomechanical parameters are used to indicate the relationship between the calorie deficit and fat breakdown rate caused by the fitness plan to the user's body. The prediction module is used to input the current body shape model and the biomechanical parameters into a pre-trained target model to obtain the user's predicted body shape model. The target model is used to analyze the influence of the biomechanical parameters on the current body shape model in order to output the predicted body shape model. The push module is used to push the predicted body shape model to the user.

[0014] Fourthly, embodiments of this application also provide a model training apparatus, the apparatus comprising: The second acquisition module is used to acquire the preset model and training dataset; The training module is used to train the preset model based on the training dataset to obtain the target model as described in the first aspect. The target model is used to analyze the influence of the input biomechanical parameters on the current body shape modeling to output a predicted body shape modeling.

[0015] Fifthly, embodiments of this application also provide an electronic device, including a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the fitness information push method as described in the first aspect, or the steps of the model training method as described in the second aspect.

[0016] Sixthly, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the fitness information push method as described in the first aspect, or the steps of the model training method as described in the second aspect.

[0017] In a seventh aspect, a computer program product is provided, including computer instructions that, when executed by a processor, implement the steps of the fitness information push method as described in the first aspect, or the steps of the model training method as described in the second aspect.

[0018] In this embodiment, the user's current body shape model and fitness plan are obtained; based on the fitness plan, biomechanical parameters are determined, which indicate the relationship between the calorie deficit and fat breakdown rate caused by the fitness plan on the user's body; the current body shape model and the biomechanical parameters are input into a pre-trained target model to obtain the user's predicted body shape model, wherein the target model is used to analyze the impact of the biomechanical parameters on the current body shape model to output the predicted body shape model; the predicted body shape model is then pushed to the user. Specifically, the fitness plan is converted into biomechanical parameters, which dynamically reflect the impact of the fitness plan on the body shape. Then, the pre-trained target model analyzes the user's current body shape model and the corresponding biomechanical parameters of the fitness plan to specifically analyze the impact of the fitness plan on the user's body shape, thereby adjusting the current body shape model to obtain the predicted body shape model. Compared with existing technologies, the obtained predicted body shape model is more accurate; that is, the method of this embodiment can improve the accuracy of pushed information. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is one of the flowcharts illustrating the fitness information push method provided in the embodiments of this application; Figure 2 This is a second flowchart illustrating the fitness information push method provided in this application embodiment; Figure 3 This is a flowchart illustrating the model training method provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of a fitness information push device provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a model training device provided in one embodiment of this application; Figure 6 This is a structural diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] like Figure 1 As shown in the embodiment of this application, the fitness information push method includes the following steps: Step 101: Obtain the user's current body shape model and fitness plan.

[0023] In this embodiment, as Figure 2 As shown, body modeling is based on three-dimensional images ( Figure 2 The fitness plan is displayed in the form of current modeling images and fitness prediction images. The fitness plan can include aerobic exercise, strength training programs, and corresponding training frequency, intensity, duration, etc., which are displayed in text format.

[0024] Step 102: Based on the fitness plan, determine biomechanical parameters, which are used to indicate the relationship between the calorie deficit and fat breakdown rate caused by the fitness plan to the user's body.

[0025] In this step, the fitness plan, presented in text form, is transformed into biomechanical parameters that allow for a more intuitive understanding of its impact on the human body. Understandably, the body model can be adjusted accordingly based on the stated calorie deficit and fat breakdown rate.

[0026] Step 103: Input the current body shape model and the biomechanical parameters into the pre-trained target model to obtain the user's predicted body shape model. The target model is used to analyze the influence of the biomechanical parameters on the current body shape model in order to output the predicted body shape model.

[0027] Through a pre-trained target model ( Figure 2 The spatiotemporal dynamic prediction model shown analyzes the user's current body shape model and the biomechanical parameters corresponding to the fitness plan, thereby analyzing the impact of the fitness plan on the user's body shape, and adjusting the current body shape model to obtain the predicted body shape model. Figure 2 This allows for a more intuitive understanding of the current body shape modeling ( Figure 2 The current modeling image shown is processed by the spatiotemporal dynamic prediction model to obtain the predicted body shape model. Figure 2 The fitness prediction image shown.

[0028] It should be noted that the target model has a time-step iteration mechanism. A time-step parameter is introduced, and at each time step, based on the current body shape and biomechanical parameters, the change in fat-muscle distribution at the next moment is predicted. This iterative process generates a continuous and natural gradual body shape change path. State transition equation: Attention(Q,K,V)=Softmax((QK^T / )⊙W_mask)V; Q / K / V correspond to the current state S_t, the historical state sequence S_(0:t-1), and the biomechanical parameter P_t, respectively; W_mask is a structured attention mask to suppress interference from invalid time steps (such as future training plans that have not been executed). It should be noted that t is a positive integer, and when t is 1, the current state and the historical state sequence are consistent.

[0029] The output generates the deformation gradient for the next time step through a feedforward network (FFN): ΔS_(t+1)=FFN(Attention(Q,K,V).

[0030] This deformation gradient accurately describes the degree and direction of possible changes in the distribution of muscle and fat in different parts of the human body from the current moment to the next moment. By continuously repeating the above process, that is, making predictions based on the current state and relevant parameters at each time step, and updating the body shape state using the generated deformation gradient, the system can gradually construct a continuous and natural body shape gradient path, thereby obtaining the final predicted body shape model.

[0031] Step 104: Push the predicted body shape model to the user.

[0032] The system pushes information including the predicted body shape model to the user.

[0033] In this embodiment, the fitness plan is converted into biomechanical parameters, which can dynamically reflect the impact of the fitness plan on the body shape. Then, a pre-trained target model is used to analyze the user's current body shape model and the biomechanical parameters corresponding to the fitness plan. Specifically, the impact of the fitness plan on the user's body shape is analyzed, thereby adjusting the current body shape model to obtain the predicted body shape model. Compared with existing technologies, the obtained predicted body shape model is more accurate. In other words, the method of this embodiment can improve the accuracy of push information.

[0034] Optionally, after obtaining the user's current body shape model and fitness plan, the method further includes: Upon receiving an adjustment operation from the user regarding the current body shape model, obtain the desired body shape model obtained based on the adjustment operation. The current body shape model, the fitness plan, and the desired body shape model are input into the target model to obtain the modified fitness plan. The target model is used to analyze the current body shape model and the desired body shape model to adjust the fitness plan and obtain the modified fitness plan. The modified fitness plan is pushed to the user.

[0035] Please see below. Figure 2 The adjustment operation of the current body shape model refers to, through... Figure 2 The virtual avatar interactive editing module shown allows users to select a portion of the body model (current modeling image) and perform operations such as zooming in and out to represent the user's desired ideal body shape, i.e., the expected body shape model.

[0036] Using the target model, the difference between the current body shape model and the desired body shape model can be analyzed to determine how much calorie deficit is needed and what fat breakdown rate needs to be achieved. Based on this, the fitness plan can be adjusted to obtain a modified fitness plan that meets the requirements.

[0037] In this embodiment, not only can users forward model their future body shape based on the fitness plan, but the fitness plan can also be reverse-optimized by modifying the virtual body to meet the user's personalized needs.

[0038] Optionally, the target model is further used for: Extract the time-series features of the fitness plan; Based on the difference between the current body shape model and the desired body shape model, the deformation characteristics are determined; The time series features and the deformation features are encoded into a unified dimension tensor to obtain the first encoded data corresponding to the time series features and the second encoded data corresponding to the deformation features; Based on preset physiological tolerance constraint parameters, the first encoded data, and the second encoded data, the fitness plan is adjusted to obtain the modified fitness plan. The physiological tolerance constraint parameters are used to ensure that the user's body shape changes conform to physiological laws.

[0039] In this embodiment, the time-series features refer to the feature plans obtained by sorting them in time sequence. For example, the biomechanical parameters of the fitness plan corresponding to day one are a1, and the biomechanical parameters of the fitness plan corresponding to day two are a2, etc. The deformation features refer to the differences in body shape between the current body shape model and the desired body shape model, such as the rate of change in waist circumference and the increase in muscle volume.

[0040] like Figure 2As shown, the target model employs a multi-head cross-modal attention mechanism, using a temporal Transformer architecture to process the time-series features of the fitness plan. Through this mechanism, it achieves efficient correlation between biomechanical parameters and body deformation. The time-series features and deformation features are encoded into a unified-dimensional tensor (both types of data must be analyzed within the same dimension), and feature interaction is achieved through a 12-head cross-modal attention layer. For example, when calculating the impact of squat training on gluteus maximus volume in week t, the attention weight matrix correlates the number of sets, load parameters, and point cloud density changes of the corresponding anatomical region.

[0041] Furthermore, the target model possesses a physical constraint mechanism, which dynamically controls human body deformation by embedding preset physiological tolerance constraint parameters (which can be mathematical equations and biomechanical models). Its core function is to ensure that virtual body changes conform to anatomical laws and suppress unreasonable deformations. For example, using 3D Gaussian sputtering technology, the point cloud density ρ(r) is mapped to the anatomical structure: | , ; K=15 corresponds to the major muscle groups, and the covariance matrix reflects the direction of muscle fibers. By constructing and adjusting the covariance matrix, the morphological changes of different muscles during contraction and relaxation, as well as the interactions and influences between muscles, can be simulated. The changes in human body shape displayed in the virtual environment can highly restore the real physiological process.

[0042] This embodiment allows for a more accurate analysis of the actual physiological processes underlying changes in human body shape, thereby improving the accuracy of the results.

[0043] In an optional implementation, the process of adjusting the fitness plan based on preset physiological tolerance constraint parameters, the first encoded data, and the second encoded data to obtain the modified fitness plan is as follows: Figure 2 The bidirectional inference engine shown can execute in the following ways: The required change in muscle / fat is calculated by inverse differentiation based on the second encoded data. Then, the required biomechanical parameters are derived based on the required change in muscle / fat. The first encoded data is then adjusted based on the derived biomechanical parameters to obtain the adjusted first encoded data. This is then used to derive a modified fitness plan. Throughout the process, the physiological tolerance constraint parameters restrict all types of parameters to prevent data that does not conform to physiological laws.

[0044] Optionally, determining biomechanical parameters based on the fitness plan includes: Convert the fitness plan into exercise parameters; Obtain a first function from a pre-defined biomechanical simulation library. This first function is used to indicate the force generated by human muscle groups at different lengths and contraction speeds. The biomechanical parameters are determined based on the motion parameters and the first function.

[0045] The method steps in this embodiment can be derived from... Figure 2 The biomechanical parameter mapping module, as shown, performs structured analysis on the input fitness plan (including details such as aerobic exercise, strength training programs, training frequency, intensity, and duration), transforming it into quantifiable exercise parameters. Different intensities of strength training are mapped to parameters such as muscle stress threshold and fatigue recovery cycle, while aerobic exercise is associated with calorie consumption rate and cardiopulmonary function improvement indicators, providing dynamic driving factors for predicting body shape changes. Integrating the OpenSim biomechanical simulation library, it obtains the force-length-velocity relationship equations (first function) for 20 muscle groups in the human body. ATP metabolic efficiency is collected in real time via flexible electrodes, and a nonlinear equation is constructed to describe the dynamic relationship between calorie deficit and fat breakdown rate, ensuring that the prediction conforms to the laws of exercise physiology.

[0046] Optionally, obtaining the user's current body shape model includes: Acquire first data and second data. The first data is three-dimensional point cloud data generated after three-dimensional scanning of the user's human body. The second data is a human body volume density model obtained after performing bioelectrical impedance analysis and human body displacement air volume analysis on the user's human body. The first data is adjusted based on the second data to obtain the current body shape model.

[0047] In this embodiment, the first data can be acquired by using a high-resolution 3D sensing camera to complete a 360° human body scan within 40 seconds. Figure 2 The optical 3D scan shown generates 3D point cloud data containing 200,000+ points (via...). Figure 2 The 3D Gaussian point cloud modeling module shown is executed, and the three-dimensional skeletal point detection technology is used to locate key anatomical landmarks such as the acromion and anterior superior iliac spine, and to divide the muscle into 15 parts, including the pectoralis major and rectus abdominis muscles.

[0048] The second data can be obtained by measuring the electrical impedance values ​​of the limbs and torso using a BIA module. Figure 2 The electrode data shown can be used to estimate the water content (muscle mass) and fat conductivity (fat mass) of muscles in various parts of the body. By accurately measuring the amount of air replaced by the human body volume, a volume density model can be established to verify the biological rationality of the optical scanning data.

[0049] The second set of data is used to verify the biological validity of the first set of data, adjust any unreasonable modeling content, and thereby improve the accuracy of the current body shape modeling.

[0050] like Figure 3 As shown in the embodiments of this application, a model training method is also provided, including the following steps: Step 201: Obtain the preset model and training dataset; Step 202: Train the preset model based on the training dataset to obtain... Figure 1 The target model in the fitness information push method of the illustrated embodiment is used to analyze the influence of the input biomechanical parameters on the input current body shape modeling, so as to output a predicted body shape model.

[0051] In this embodiment, a preset model is trained using a training dataset to obtain a target model capable of analyzing the impact of input biomechanical parameters on the modeling of the current body shape, thereby outputting a predictive model for body shape modeling. This target model is then applied... Figure 1 The fitness information push method shown can improve the accuracy of predictive body shape modeling.

[0052] Optionally, the preset model includes a first sub-model, and the training dataset includes a first subset, which includes historical current body shape modeling, historical biomechanical parameters, and real body shape modeling corresponding to the historical current body shape modeling and the historical biomechanical parameters. The step of training the preset model based on the training dataset to obtain the target model includes: Using the historical current body shape model and the historical biomechanical parameters as input data, and the real body shape model as the label, the first sub-model is trained to obtain a body shape dynamic prediction sub-model. The first loss function during the training of the first sub-model is determined based on the real body shape model and the predicted body shape model output by the first sub-model. The body shape dynamic prediction sub-model is used to analyze the influence of the input biomechanical parameters on the input current body shape model in order to output a predicted body shape model.

[0053] In this embodiment, the first loss function is: , in, It is used to measure the difference between predicted body size and actual body size, and is calculated using indicators such as mean squared error; Used to constrain topological changes during the gradual transformation of body shape, ensuring that the changes in body shape conform to biomechanical principles; This method is used to assess the match between fitness plans and user needs, and is calculated based on a reward function. The model parameters are updated using a backpropagation algorithm to minimize the loss function, thus achieving optimized model training.

[0054] Optionally, the preset model includes a second sub-model, and the training dataset includes a second subset, which includes historical current body shape modeling, historical fitness plans, historical expected body shape modeling, and real fitness plans corresponding to the historical current body shape modeling, the historical fitness plans, and the historical expected body shape modeling. The step of training the preset model based on the training dataset to obtain the target model includes: The historical current body shape model, the historical fitness plan, and the historical desired body shape model are used as input data, and the actual fitness plan is used as a label to train the second sub-model to obtain a fitness plan adjustment sub-model. The fitness plan adjustment sub-model is used to analyze the input current body shape model and the input desired body shape model in order to adjust the input fitness plan and obtain a modified fitness plan. The second loss function during the training of the second sub-model is determined based on the real fitness plan and the modified fitness plan output by the second sub-model.

[0055] During the training of the first and second sub-models, the training data needs to be cleaned and preprocessed to remove outliers and noisy data. The training dataset is divided into a 70% training set, 15% validation set, and 15% test set for model training, validation, and testing. After each training cycle, the model performance is evaluated using the validation set data, and model hyperparameters, such as the learning rate and weight coefficients, are adjusted to improve the model's generalization ability. After training is complete, the model is finally evaluated using the test set data to ensure that the model can accurately predict body shape changes and generate reasonable fitness plans under different user data and scenarios.

[0056] See Figure 4 , Figure 4 This is a structural diagram of a fitness information push device provided in one embodiment of this application. Figure 4 As shown, the device 300 includes: The first acquisition module 301 is used to acquire the user's current body shape model and fitness plan; The first determining module 302 is used to determine biomechanical parameters based on the fitness plan, wherein the biomechanical parameters are used to indicate the relationship between the calorie deficit and fat breakdown rate caused by the fitness plan to the user's body. The prediction module 303 is used to input the current body shape model and the biomechanical parameters into a pre-trained target model to obtain the user's predicted body shape model. The target model is used to analyze the influence of the biomechanical parameters on the current body shape model in order to output the predicted body shape model. The push module 304 is used to push the predicted body shape model to the user.

[0057] Optionally, the device 300 is also used for: Upon receiving an adjustment operation from the user regarding the current body shape model, obtain the desired body shape model obtained based on the adjustment operation. The current body shape model, the fitness plan, and the desired body shape model are input into the target model to obtain the modified fitness plan. The target model is used to analyze the current body shape model and the desired body shape model to adjust the fitness plan and obtain the modified fitness plan. The modified fitness plan is pushed to the user.

[0058] Optionally, the target model is further used for: Extract the time-series features of the fitness plan; Based on the difference between the current body shape model and the desired body shape model, the deformation characteristics are determined; The time series features and the deformation features are encoded into a unified dimension tensor to obtain the first encoded data corresponding to the time series features and the second encoded data corresponding to the deformation features; Based on preset physiological tolerance constraint parameters, the first encoded data, and the second encoded data, the fitness plan is adjusted to obtain the modified fitness plan. The physiological tolerance constraint parameters are used to ensure that the user's body shape changes conform to physiological laws.

[0059] Optionally, the first determining module 302 is further configured to: Convert the fitness plan into exercise parameters; Obtain a first function from a pre-defined biomechanical simulation library. This first function is used to indicate the force generated by human muscle groups at different lengths and contraction speeds. The biomechanical parameters are determined based on the motion parameters and the first function.

[0060] Optionally, the first acquisition module 301 is further configured to: Acquire first data and second data. The first data is three-dimensional point cloud data generated after three-dimensional scanning of the user's human body. The second data is a human body volume density model obtained after performing bioelectrical impedance analysis and human body displacement air volume analysis on the user's human body. The first data is adjusted based on the second data to obtain the current body shape model.

[0061] The fitness information push device 300 provided in this application embodiment is capable of implementing... Figure 1 The entire technical process of the fitness information push method shown is to achieve the same technical effect, and will not be described again here to avoid repetition.

[0062] See Figure 5 , Figure 5 This is a structural diagram of a model training device provided in an embodiment of this application. Figure 5 As shown, the device 400 includes: The second acquisition module 401 is used to acquire the preset model and training dataset; Training module 402 is used to train the preset model based on the training dataset to obtain... Figure 1 The target model in the fitness information push method of the illustrated embodiment is used to analyze the influence of the input biomechanical parameters on the input current body shape modeling, so as to output a predicted body shape model.

[0063] Optionally, the preset model includes a first sub-model, and the training dataset includes a first subset, which includes historical current body shape modeling, historical biomechanical parameters, and real body shape modeling corresponding to the historical current body shape modeling and the historical biomechanical parameters; the training module 402 includes a first training sub-module, used for: Using the historical current body shape model and the historical biomechanical parameters as input data, and the real body shape model as the label, the first sub-model is trained to obtain a body shape dynamic prediction sub-model. The first loss function during the training of the first sub-model is determined based on the real body shape model and the predicted body shape model output by the first sub-model. The body shape dynamic prediction sub-model is used to analyze the influence of the input biomechanical parameters on the input current body shape model in order to output a predicted body shape model.

[0064] Optionally, the preset model includes a second sub-model, and the training dataset includes a second subset, which includes historical current body shape modeling, historical fitness plans, historical expected body shape modeling, and real fitness plans corresponding to the historical current body shape modeling, the historical fitness plans, and the historical expected body shape modeling. Training module 402 includes a second training submodule, used for: The historical current body shape model, the historical fitness plan, and the historical desired body shape model are used as input data, and the actual fitness plan is used as a label to train the second sub-model to obtain a fitness plan adjustment sub-model. The fitness plan adjustment sub-model is used to analyze the input current body shape model and the input desired body shape model in order to adjust the input fitness plan and obtain a modified fitness plan. The second loss function during the training of the second sub-model is determined based on the real fitness plan and the modified fitness plan output by the second sub-model.

[0065] The model training device 400 provided in this application embodiment is capable of implementing... Figure 3 The entire technical process of the model training method shown is used to achieve the same technical effect, and will not be repeated here to avoid duplication.

[0066] The fitness information push device 300 and model training device 400 in this application embodiment can be electronic devices or components within electronic devices, such as integrated circuits or chips. The electronic device can be a terminal or other devices besides a terminal. For example, the electronic device can be a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc. Non-mobile electronic devices can also be servers, network attached storage (NAS), personal computers (PCs), televisions (TVs), ATMs, or self-service machines, etc. This application embodiment does not specifically limit the specific devices.

[0067] Optionally, such as Figure 6 As shown, this application embodiment also provides an electronic device 500, including a processor 501 and a memory 502. The memory 502 stores a program or instructions that can run on the processor 501. When the program or instructions are executed by the processor 501, they implement the various steps of the above-described fitness information push method or model training method embodiment and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0068] It should be noted that the electronic devices in the embodiments of this application include the mobile electronic devices and non-mobile electronic devices described above.

[0069] This application also provides a computer-readable storage medium storing a computer program. When executed by a processor, this computer program implements the various processes of the above-described fitness information push method or model training method embodiments, and achieves the same technical effect. To avoid repetition, it will not be described again here. The computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0070] This application also provides a computer program product, including computer instructions. When executed by a processor, these computer instructions implement the various processes of the above-described fitness information push method or model training method embodiments and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0071] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0072] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0073] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A fitness information push method, characterized by, The method comprises: obtaining current body modeling and a fitness plan of a user; determining biomechanical parameters based on the fitness plan, the biomechanical parameters being used to indicate the relationship between the heat gap and the fat decomposition rate caused by the fitness plan on the human body of the user; inputting the current body modeling and the biomechanical parameters into a pre-trained target model to obtain predicted body modeling of the user, wherein the target model is used to analyze the influence of the biomechanical parameters on the current body modeling to output predicted body modeling; pushing the predicted body modeling to the user.

2. The method of claim 1, wherein, After obtaining the current body modeling and the fitness plan of the user, the method further comprises: in a case where an adjustment operation of the user on the current body modeling is received, obtaining expected body modeling obtained by adjusting the current body modeling based on the adjustment operation; inputting the current body modeling, the fitness plan and the expected body modeling into the target model to obtain a modified fitness plan, wherein the target model is used to analyze the current body modeling and the expected body modeling to adjust the fitness plan to obtain the modified fitness plan; pushing the modified fitness plan to the user.

3. The method of claim 2, wherein, The target model is further used to: extract time sequence features of the fitness plan; determine deformation features based on the difference between the current body modeling and the expected body modeling; encode the time sequence features and the deformation features into a unified dimensional tensor to obtain first encoding data corresponding to the time sequence features and second encoding data corresponding to the deformation features; adjust the fitness plan based on a preset physiological tolerance constraint parameter, the first encoding data and the second encoding data to obtain the modified fitness plan, the physiological tolerance constraint parameter being used to ensure that the body shape change of the user conforms to physiological laws.

4. The method of claim 1, wherein, The determination of the biomechanical parameters based on the fitness plan comprises: converting the fitness plan into motion parameters; obtaining a first function from a preset biomechanical simulation library, the first function being used to indicate the force generated by human muscle groups at different lengths and contraction speeds; determining the biomechanical parameters based on the motion parameters and the first function.

5. The method according to any one of claims 1 to 4, characterized in that, The obtaining of the current body modeling of the user comprises: obtaining first data and second data, the first data being three-dimensional point cloud data generated after three-dimensional scanning of the human body of the user, and the second data being a human body volume density model obtained after bioelectrical impedance analysis and human body air displacement analysis of the human body of the user; adjusting the first data based on the second data to obtain the current body modeling.

6. A model training method, comprising: The method comprises: obtaining a preset model and a training data set; training the preset model based on the training data set to obtain the target model of any one of claims 1 to 5, the target model being used to analyze the influence of the input biomechanical parameters on the input current body modeling to output predicted body modeling.

7. The method of claim 6, wherein, The preset model comprises a first sub-model, and the training data set comprises a first subset, and the first subset comprises historical current body shape modeling, historical biomechanical parameters, and real body shape modeling corresponding to the historical current body shape modeling and the historical biomechanical parameters; The training of the preset model based on the training data set comprises: The historical current body shape modeling and the historical biomechanical parameters are taken as input data, and the real body shape modeling is taken as a label, and the first sub-model is trained to obtain a body shape dynamic prediction sub-model, wherein a first loss function of the first sub-model during the training is determined based on the real body shape modeling and predicted body shape modeling output by the first sub-model, and the body shape dynamic prediction sub-model is used for analyzing the influence of input biomechanical parameters on input current body shape modeling to output predicted body shape modeling.

8. The method of claim 6, wherein, The preset model comprises a second sub-model, and the training data set comprises a second subset, and the second subset comprises historical current body shape modeling, historical fitness plans, historical expected body shape modeling, and real fitness plans corresponding to the historical current body shape modeling, the historical fitness plans, and the historical expected body shape modeling; The training of the preset model based on the training data set comprises: The historical current body shape modeling, the historical fitness plans, and the historical expected body shape modeling are taken as input data, and the real fitness plans are taken as a label, and the second sub-model is trained to obtain a fitness plan adjustment sub-model, wherein the fitness plan adjustment sub-model is used for analyzing input current body shape modeling and input expected body shape modeling to adjust input fitness plans to obtain modified fitness plans; A second loss function of the second sub-model during the training is determined based on the real fitness plans and the modified fitness plans output by the second sub-model.

9. A health information push device, characterized by comprising: The device comprises: A first obtaining module is configured to obtain current body shape modeling and a fitness plan of a user; A first determining module is configured to determine biomechanical parameters based on the fitness plan, wherein the biomechanical parameters are used to indicate the relationship between the heat gap and the fat decomposition rate of the human body caused by the fitness plan; A prediction module is configured to input the current body shape modeling and the biomechanical parameters into a pre-trained target model to obtain predicted body shape modeling of the user, wherein the target model is used to analyze the influence of the biomechanical parameters on the current body shape modeling to output predicted body shape modeling; A pushing module is configured to push the predicted body shape modeling to the user.

10. A model training apparatus, comprising: The device comprises: A second obtaining module is configured to obtain a preset model and a training data set; A training module is configured to train the preset model based on the training data set to obtain the target model of any one of claims 1 to 5, wherein the target model is used to analyze the influence of input biomechanical parameters on input current body shape modeling to output predicted body shape modeling.

11. An electronic device, comprising: The computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the steps of the fitness information pushing method according to any one of claims 1 to 5, or the steps of the model training method according to any one of claims 6 to 8.

12. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the steps of the fitness information pushing method according to any one of claims 1 to 5, or the steps of the model training method according to any one of claims 6 to 8.

13. A computer program product, characterised in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the steps of the fitness information pushing method according to any one of claims 1 to 5, or the steps of the model training method according to any one of claims 6 to 8.

Images