Strength training action intelligent calculation evaluation system based on multi-modal fusion

Abstract

The invention provides a strength training action intelligent calculation evaluation system based on multi-modal fusion, which is used for executing a strength training action evaluation method, and comprises the following steps: collecting multi-modal input data in a strength training process of a trainer; the input data comprises visual information, muscle neural signals and physiological data of the trainee, the visual information comprises limb actions and posture tracks, and the physiological data comprises heart rate and blood sugar; preprocessing the input data, and inputting the preprocessed input data into a strength training expert agent, so that the strength training expert agent outputs an evaluation result of the trainer in combination with the cue word and the strength training expert knowledge base; and displaying an evaluation result, and determining a training plan of the trainer based on the evaluation result. The physical strength training auxiliary logic is comprehensively upgraded from data basis, intelligent analysis to training closed loop, more systematic technical support is provided for scientific training and efficient improvement, and the requirements of modern physical strength training for precision, individuation and long-term effect are met.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a strength training action intelligent calculation evaluation system based on multimodal fusion. Background Art

[0002] Currently, artificial intelligence (AI) is developing rapidly, and it is foreseeable that many jobs will see significant reductions in labor costs or even be directly replaced by AI. Currently, with people's growing awareness of health, strength training has become widely accepted as an important means to improve physical fitness and body shape, and the process of strength training cannot be replaced by AI in any way. However, without the guidance of professional coaches, independent training can lead to low training efficiency, improper training movements, and even the risk of injury. Traditional training guidance relies on personal trainers, but the level of trainers varies greatly, making it difficult for beginners to identify them. Hiring personal trainers is also generally expensive, making it difficult for everyone to access them.

[0003] In addition, existing AI coaching systems only rely on single-modal data. For example, the use of human motion recognition technology based on target detection can only provide rough posture scores and cannot form interactive, personalized and long-term effective training guidance plans. In particular, they lack the ability to analyze deep data such as physical condition, neural recruitment, and muscle load. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a strength training action intelligent evaluation system based on multimodal fusion, so as to construct an expert intelligent body to evaluate strength training actions, comprehensively upgrade the strength training auxiliary logic from data basis, intelligent analysis to training closed loop, and provide more systematic technical support for scientific training and efficient improvement, and adapt to the modern strength training needs for precision, personalization and long-term effectiveness.

[0005] In a first aspect, an embodiment of the present invention provides a strength training action evaluation method for executing a strength training action evaluation method, the strength training action evaluation method comprising: collecting multimodal input data during a trainee's strength training process; wherein the input data comprises: visual information, muscle nerve signals and physiological data of the trainee, the visual information comprises: limb movements and posture trajectories, and the physiological data comprises: heart rate and blood sugar; preprocessing the input data, and inputting the preprocessed input data into a strength training expert intelligent agent, so that the strength training expert intelligent agent outputs the trainee's evaluation results in combination with prompt words and a strength training expert knowledge base; displaying the evaluation results, and determining the trainee's training plan based on the evaluation results.

[0006] In an optional embodiment of the present application, the above-mentioned step of collecting multimodal input data includes: capturing visual information through a camera; obtaining muscle nerve signals through an electromyography sensor; and obtaining physiological data through a detection bracelet.

[0007] In an optional embodiment of the present application, the above-mentioned step of determining the trainee's training plan based on the evaluation results includes: judging whether the evaluation results meet the preset training requirements; if so, generating maintenance suggestions and a training plan for the next stage; if not, generating corrective suggestions and a training plan for the unmet training requirements.

[0008] In an optional embodiment of the present application, the above method also includes: obtaining historical input data, performing data cleaning and preprocessing on the historical input data; inputting the historical input data after data cleaning and preprocessing into the multimodal large model for fine-tuning training; constructing an expert knowledge base based on the historical input data after data cleaning and preprocessing; and combining prompt words to construct a strength training expert knowledge base based on the multimodal large model after fine-tuning training and the constructed expert knowledge base.

[0009] In an optional embodiment of the present application, after the above-mentioned steps of data cleaning and preprocessing of historical input data, the method further includes: performing action posture recognition on historical visual information, and performing manual correction and data labeling; extracting characteristic values ​​corresponding to the target action based on historical muscle nerve signals; and manually labeling the historical physiological data of key nodes; wherein the key nodes include: when the target muscle begins to exert force, when the target muscle is exhausted, and when the target muscle group is resting.

[0010] In an optional embodiment of the present application, the above-mentioned step of inputting the historical input data after data cleaning and preprocessing into the multimodal large model for fine-tuning training includes: inputting the characteristic values ​​of historical visual information and historical muscle nerve signals into the multimodal large model for fine-tuning training.

[0011] In an optional embodiment of the present application, the above-mentioned step of constructing an expert knowledge base based on the historical input data after data cleaning and preprocessing includes: constructing an expert knowledge base based on the historical physiological data of key nodes.

[0012] In the second aspect, an embodiment of the present invention also provides a strength training action evaluation system. The strength training action intelligent evaluation system based on multimodal fusion includes: an input data acquisition module, which is used to collect multimodal input data during the trainee's strength training; wherein the input data includes: the trainee's visual information, muscle nerve signals and physiological data, the visual information includes: limb movements and posture trajectories, and the physiological data includes: heart rate and blood sugar; a strength training expert intelligent agent processing module, which is used to pre-process the input data and input the pre-processed input data into the strength training expert intelligent agent, so that the strength training expert intelligent agent outputs the trainee's evaluation results in combination with the prompt words and the strength training expert knowledge base; an evaluation result feedback module, which is used to display the evaluation results and determine the trainee's training plan based on the evaluation results.

[0013] In an optional embodiment of the present application, the above-mentioned strength training action intelligent evaluation system based on multimodal fusion also includes: a strength training expert knowledge base construction module, which is used to construct an expert knowledge base based on historical input data after data cleaning and preprocessing; combined with prompt words, a strength training expert knowledge base is constructed based on the multimodal large model after fine-tuning training and the constructed expert knowledge base.

[0014] In an optional embodiment of the present application, the above-mentioned strength training expert knowledge base construction module is also used to recognize action postures of historical visual information, and perform manual correction and data labeling; extract characteristic values ​​corresponding to target actions based on historical muscle nerve signals; and manually label historical physiological data of key nodes; wherein, key nodes include: when the target muscle begins to exert force, when the target muscle is exhausted, and when the target muscle group is resting.

[0015] The embodiments of the present invention bring the following beneficial effects: An embodiment of the present invention provides a multimodal fusion-based intelligent evaluation system for strength training movements. The system collects multimodal input data during a trainee's strength training. The input data includes visual information, muscle nerve signals, and physiological data of the trainee. Visual information includes limb movements and posture trajectories, and physiological data includes heart rate and blood sugar. The system preprocesses the input data and feeds it into a strength training expert agent, which then outputs an evaluation result of the trainee based on prompts and a knowledge base. The evaluation result is then displayed, and a training plan is determined based on the evaluation result. This approach overcomes the limitations of single-dimensional cognition through multimodal deep data collection. Expert agents are relied upon to achieve intelligent and precise reasoning feedback, replacing experience and simple rules. A continuous closed-loop optimization mechanism is established to promote dynamic iteration of training results. Compared to existing technologies, this system comprehensively upgrades the auxiliary logic of strength training, from data foundation and intelligent analysis to training closed loop, providing more systematic technical support for scientific training and efficient improvement, and adapting to the modern strength training requirements for precision, personalization, and long-term effectiveness.

[0016] Other features and advantages of the present disclosure will be set forth in the following description, or some features and advantages may be inferred or unambiguously determined from the description, or may be learned by practicing the above-mentioned technology of the present disclosure.

[0017] In order to make the above-mentioned objectives, features and advantages of the present disclosure more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 A flow chart of a strength training action evaluation method provided by an embodiment of the present invention; Figure 2 A schematic diagram of the overall design of a strength training evaluation system provided by an embodiment of the present invention; Figure 3 A flowchart of a method for constructing a strength training expert agent provided by an embodiment of the present invention; Figure 4 A schematic diagram of a method for constructing a strength training expert agent provided by an embodiment of the present invention; Figure 5A flowchart of another strength training movement evaluation method provided by an embodiment of the present invention; Figure 6 A schematic diagram of a strength training action evaluation method provided by an embodiment of the present invention; Figure 7 A schematic diagram of the structure of a strength training action intelligent evaluation system based on multimodal fusion provided by an embodiment of the present invention; Figure 8 A schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0021] Currently, existing AI coaching systems have the following flaws: 1. Human resource dependence and cost issues: Traditional strength training guidance is highly dependent on personal trainers. However, the level of coaches varies widely, and beginners find it difficult to identify their professional abilities. In addition, hiring personal trainers is expensive, making it difficult to popularize training guidance. As a result, a large number of fitness enthusiasts are unable to obtain continuous and professional training guidance. 2. Single data modality and rough evaluation: Most existing AI coaching systems rely on single-modality data, such as human motion recognition technology based on target detection. They can only roughly score human posture based on appearance and cannot fully analyze the trainee's movement details and movement patterns.

[0022] 3. Lack of personalized and long-term effective guidance: Existing systems based on single-modality data make it difficult to develop personalized training guidance plans based on individual trainees' physical conditions, fitness levels, training goals, etc., and are also unable to dynamically adjust training plans as trainees progress, making it difficult to achieve long-term and effective training guidance.

[0023] 4. Inability to obtain deep data: Existing technologies lack the ability to analyze deep data such as physical condition, neural recruitment, and muscle load. They are unable to fully understand the trainees’ physiological reactions and muscle working status during training, and it is difficult to provide trainees with scientific training suggestions from a physiological perspective.

[0024] Based on this, an embodiment of the present invention provides a strength training action intelligent evaluation system based on multimodal fusion, specifically providing a strength training action evaluation system based on multimodal fusion, which mainly includes the overall design of the system, the construction of a strength training expert intelligent body, and the evaluation process of the strength training expert intelligent body.

[0025] The method provided by the embodiments of the present invention can significantly reduce labor costs by building an automated strength training movement assessment system and using artificial intelligence technology to replace some manual guidance work, allowing more people to enjoy professional-level training guidance services at a lower cost. It can also use multimodal fusion technology to integrate multiple different types of data, such as visual data, electromyographic signal data, inertial sensor data, etc., to conduct in-depth analysis of training movements from multiple dimensions, providing more accurate and comprehensive movement assessment results than single-modal data. Based on the comprehensive analysis of multimodal data, combined with deep data such as the trainee's physical condition, neural recruitment, and muscle load, it can deeply understand the trainee's individual characteristics and training needs, tailor a personalized training guidance plan for each trainee, and continuously optimize it based on data changes during training to achieve long-term and effective training guidance. It can also obtain and process this deep data through the collection and fusion analysis of multimodal data, provide trainees with scientific training suggestions based on physiological levels, help trainees perform strength training more safely and efficiently, reduce the risk of injury due to improper training, and improve training effectiveness.

[0026] To facilitate understanding of this embodiment, a strength training action evaluation method disclosed in an embodiment of the present invention is first introduced in detail.

[0027] Example 1: The embodiment of the present invention provides a strength training action intelligent evaluation system based on multimodal fusion, which is used to perform the strength training action evaluation method. Figure 1 The flowchart of a strength training action evaluation method shown in FIG. 1 includes the following steps: Step S102: collecting multimodal input data during the trainee's strength training.

[0028] See also Figure 2 The overall design diagram of a strength training evaluation system shown in FIG. 4 may include: a data acquisition module, a multimodal expert agent server, and a screen feedback terminal.

[0029] like Figure 2 As shown, the data acquisition module serves as the system information input source, which can connect to the trainee's strength training process and collect multimodal input data.

[0030] The input data includes: the trainee's visual information, muscle nerve signals and physiological data. The visual information includes: limb movements and posture trajectories, and the physiological data includes: heart rate and blood sugar.

[0031] In some embodiments, the trainee's body movements, posture trajectories and other visual information can be captured by cameras, muscle nerve signals can be obtained by electromyography sensors, and physiological data such as heart rate and blood sugar can be obtained by electrocardiogram and blood sugar bracelets. The three together provide input data for the expert intelligent agent's reasoning.

[0032] Step S104, pre-processing the input data, and inputting the pre-processed input data into the strength training expert agent, so that the strength training expert agent outputs the evaluation results of the trainee in combination with the prompt words and the strength training expert knowledge base.

[0033] like Figure 2 As shown in the figure, the multimodal expert agent server serves as the core processing unit of the system. It can receive the multimodal input data transmitted by the data acquisition module. After data preprocessing, it combines the prompt words and the strength training expert knowledge base to generate structured feedback information for the strength training expert agent as the evaluation result of the trainee.

[0034] Step S106: display the evaluation results and determine the trainee's training plan based on the evaluation results.

[0035] like Figure 2 As shown, the screen feedback end is responsible for the result output function. It can receive the evaluation results of the multimodal large model server and deliver them to the screen in real time in an intuitive form (text, animation, warning signs, etc.), and provide the trainee with feedback on movement adjustment suggestions, intensity ranges, etc., so that the trainee can obtain guidance in time and adjust the training movements and intensity.

[0036] An embodiment of the present invention provides a strength training movement evaluation system for executing a strength training movement evaluation method. The system collects multimodal input data during a trainee's strength training. The input data includes visual information, muscle nerve signals, and physiological data of the trainee. The visual information includes limb movements and posture trajectories, and the physiological data includes heart rate and blood glucose. The system preprocesses the input data and feeds it into a strength training expert agent, which then outputs an evaluation result of the trainee based on prompts and a strength training expert knowledge base. The evaluation result is then displayed, and a training plan is determined based on the evaluation result. This approach overcomes the limitations of single-dimensional cognition through multimodal deep data collection. The system relies on expert agents to implement intelligent and precise reasoning feedback, replacing experience and simple rules. A continuous closed-loop optimization mechanism is established to promote dynamic iteration of training results. Compared to existing technologies, this system comprehensively upgrades the strength training auxiliary logic from data foundation, intelligent analysis, to training closed loop, providing more systematic technical support for scientific training and efficient improvement, adapting to the modern strength training requirements of precision, personalization, and long-term effectiveness.

[0037] Example 2: This embodiment provides another strength training action intelligent evaluation system based on multimodal fusion. This method is implemented on the basis of the above embodiment and focuses on describing the method of constructing a strength training expert intelligent agent. Figure 3 The flowchart of a method for constructing a strength training expert agent is shown. The method for constructing a strength training expert agent includes the following steps: Step S302: Acquire historical input data, and perform data cleaning and preprocessing on the historical input data.

[0038] See also Figure 4 The diagram shows a method for constructing a strength training expert intelligent agent. In this embodiment, strength training experts with different styles can be hired to carry out training movement data collection, using high-definition cameras to collect visual data, using electromyography sensors to collect neural signals, and using electrocardiogram and blood sugar bracelets to complete the collection of physical indicator data such as heart rate and blood sugar, and obtain historical input data.

[0039] In some embodiments, historical visual information can also be used to identify action postures, perform manual corrections, and label data; extract characteristic values ​​corresponding to target actions based on historical muscle nerve signals; and manually label historical physiological data of key nodes; key nodes include: when the target muscle begins to exert force, when the target muscle is exhausted, and when the target muscle group is resting.

[0040] like Figure 4As shown, this embodiment also performs data cleaning and preprocessing on historical input data. Data cleaning and preprocessing are performed on collected multimodal data, including visual information, neural signals, heart rate, and blood glucose levels. For the strength training expert's action images (i.e., historical visual information), Openpose (a two-stream architecture based on a convolutional neural network) can be used to perform posture recognition to obtain raw images, which are then manually corrected and labeled. For the strength training expert's muscle nerve signals (i.e., historical muscle nerve signals), data cleaning and preprocessing are performed to extract feature values ​​corresponding to the target action. Physical indicators such as blood glucose and heart rate (i.e., historical physiological data) are manually labeled, primarily to mark changes in heart rate and blood glucose levels at key points, including when the target muscle begins to exert force, when the target muscle reaches exhaustion, and when the target muscle group rests.

[0041] Step S304: input the historical input data after data cleaning and preprocessing into the multimodal large model for fine-tuning training.

[0042] In this embodiment, historical input data can be input into the multimodal large model for fine-tuning training. In some embodiments, feature values ​​of historical visual information and historical muscle nerve signals can be input into the multimodal large model for fine-tuning training.

[0043] like Figure 4 As shown, this embodiment can inject the feature value dataset of action posture visual images and muscle nerve signals into the large model for fine-tuning training. This embodiment can use the open source and intelligent deepsee-kR1 large model. There are many methods for fine-tuning training, such as Fine-tuning, Low-Rank Adaptation, PrefixTuning, etc. The specific selection shall be based on the actual effect.

[0044] Step S306: construct an expert knowledge base based on the historical input data after data cleaning and preprocessing.

[0045] This embodiment can also construct an expert knowledge base based on historical input data. In some embodiments, the expert knowledge base can be constructed based on historical physiological data of key nodes.

[0046] like Figure 4 As shown, this embodiment can construct an expert knowledge base based on the heart rate and blood glucose value data at key nodes, which serves as a guide in the subsequent strength training expert intelligent agent construction and reasoning process.

[0047] Step S308: In combination with the prompt words, a strength training expert knowledge base is constructed based on the multimodal large model after fine-tuning training and the constructed expert knowledge base.

[0048] like Figure 4 As shown, this embodiment can also construct a strength training expert intelligent agent based on the fine-tuned multimodal large model and the expert knowledge base indicating the heart rate and blood glucose value data at key nodes, combined with appropriate prompt words. This is also the core part of the entire strength training movement evaluation system.

[0049] Example 3: This embodiment provides another strength training action intelligent evaluation system based on multimodal fusion. This method is implemented on the basis of the above embodiment, focusing on the specific process of strength training action evaluation. Figure 5 A flow chart of another strength training action evaluation method is shown, and the strength training action evaluation method includes the following steps: Step S502: collecting multimodal input data during the trainee's strength training.

[0050] See also Figure 6 The schematic diagram of a strength training movement evaluation method shown in the figure shows that this embodiment can perform multimodal data collection: three types of data collection are carried out in parallel, namely, visual information collection through a camera, muscle nerve signal collection with the help of an electromyography sensor, and physiological data collection using an electrocardiogram and blood sugar bracelet, thereby obtaining all-round data of the trainee.

[0051] Step S504, pre-processing the input data, and inputting the pre-processed input data into the strength training expert agent, so that the strength training expert agent outputs the evaluation results of the trainee in combination with the prompt words and the strength training expert knowledge base.

[0052] like Figure 6 As shown, this embodiment can perform pre-processing operations such as cleaning and regularization on the collected multi-dimensional data such as visual information, muscle nerve signals and physiological data to prepare for subsequent analysis.

[0053] like Figure 6 As shown, this embodiment can also input the pre-processed input data into the strength training expert agent, and the strength training expert agent analyzes and infers the input data based on the built-in model and knowledge.

[0054] Step S506, display the evaluation results and determine whether the evaluation results meet the preset training requirements; if so, generate maintenance suggestions and a training plan for the next stage; if not, generate correction suggestions and a training plan for the unmet training requirements.

[0055] like Figure 6As shown, this embodiment can determine whether the evaluation results meet the standards (i.e., whether they meet the preset training requirements) based on the expert agent's reasoning results. If the evaluation meets the standards (i.e., meets the preset training requirements), maintenance recommendations and a training plan for the next stage can be generated. If the evaluation does not meet the standards (i.e., does not meet the preset training requirements), corrective recommendations and a training plan for the unmet training requirements can be generated.

[0056] like Figure 6 As shown, this embodiment can also provide feedback to users on generated maintenance or correction suggestions, training plans, and other content to guide their training. Users can interact with the large model in real time, thereby gaining more in-depth knowledge about strength training. After receiving feedback, the user executes the new training plan. This can then enter a continuous data collection and evaluation cycle, repeating the aforementioned data collection, preprocessing, inference, and evaluation processes to dynamically track training results.

[0057] In summary, the above-mentioned strength training action intelligent evaluation system based on multimodal fusion provided by the embodiment of the present invention can build a closed loop of "acquisition-processing-analysis-feedback-execution-loop" around the strength training process, and use multimodal data and expert intelligent agents to achieve accurate guidance and dynamic optimization of strength training.

[0058] The above-mentioned strength training action intelligent evaluation system based on multimodal fusion provided by the embodiment of the present invention has the following advantages: 1. Data Collection Dimensions and Depth Advantages: Existing strength training assistance technologies often focus on single-dimensional data, such as relying solely on visual cameras to capture movement or simply collecting physiological data through simple heart rate devices. The method provided in this embodiment builds a multimodal data collection system that encompasses visual data (camera), neural signals (electromyography sensors), and physical indicators (ECG and blood glucose wristbands). From the neural signal dimension, muscle nerve activity data collected by electromyography sensors can accurately reflect force patterns, muscle activation sequence, and intensity, supplementing the "essence of force" that visual movement observation cannot capture. Among physical indicator data, parameters such as ECG and blood glucose can characterize the trainee's physiological load limit and metabolic state. The synergistic effect of multi-dimensional data enables the system to deepen its understanding of training status from "movement appearance" to the "physiological-mechanical synergy essence," providing a more comprehensive and accurate foundation for subsequent analysis, which is difficult to achieve with single-dimensional data technologies.

[0059] 2. Advantages of Intelligent Reasoning and Feedback Mechanisms: Traditional strength training assistance relies on either coaching experience (highly subjective and difficult to standardize) or simple rule-based engines (lacking intelligent learning and adaptation capabilities). The method provided in this embodiment introduces an expert agent reasoning process, analyzing multimodal preprocessed data with built-in models and professional knowledge. After reasoning, a decision branch based on whether the assessment results meet the standards generates targeted maintenance or correction recommendations. Compared to manual experience, expert agents can exploit complex correlations between data (such as the potential connection between abnormal posture and neural signals and physiological indicators), achieving standardized and intelligent evaluation. Compared to simple rule-based engines, they possess learning and adaptation capabilities, addressing individual differences among trainees (such as physical foundation and body function) and dynamically adjusting recommendation strategies. This mechanism elevates training guidance from "experience-driven" or "simple rule-driven" to "data- and intelligent model-driven," enhancing the scientific nature and personalization of recommendations.

[0060] 3. Advantages of closed-loop training optimization: The training assistance of existing technologies is mostly single data collection and feedback, lacking continuous closed-loop optimization. The above method provided in this embodiment designs a "continuous data collection and evaluation cycle". After the user executes a new training plan, the system repeats the collection, processing, reasoning, and evaluation process. From the perspective of the training cycle, the initial training diagnoses problems and provides solutions through multimodal data; in subsequent cycles, the system continuously tracks the execution effect of the solution (such as whether the neural signal tends to be reasonable after the action is corrected, and whether the physiological indicators adapt to the new intensity), and dynamically adjusts the suggestions. This forms a closed loop of "collection-analysis-feedback-execution-recollection", allowing training optimization to change from "single static adjustment" to "continuous dynamic iteration", which is in line with the law that strength training requires long-term and gradual improvement, and ensures the steady progress of training effects.

[0061] Example 4: Corresponding to the above method embodiment, the embodiment of the present invention provides a strength training action intelligent evaluation system based on multimodal fusion, see Figure 7 The structure diagram of a strength training action intelligent evaluation system based on multimodal fusion is shown, and the strength training action intelligent evaluation system based on multimodal fusion includes: The input data acquisition module 71 is used to collect multimodal input data during the trainee's strength training. The input data includes visual information, muscle nerve signals, and physiological data of the trainee. The visual information includes body movements and posture trajectories, and the physiological data includes heart rate and blood sugar. The strength training expert agent processing module 72 is used to pre-process the input data and input the pre-processed input data into the strength training expert agent so that the strength training expert agent can output the trainee's evaluation results by combining the prompt words and the strength training expert knowledge base; The evaluation result feedback module 73 is used to display the evaluation results and determine the trainee's training plan based on the evaluation results.

[0062] An embodiment of the present invention provides an intelligent evaluation system for strength training movements based on multimodal fusion. The system includes an input data acquisition module for collecting multimodal input data during a trainee's strength training. The input data includes visual information, muscle nerve signals, and physiological data of the trainee. Visual information includes limb movements and posture trajectories, and physiological data includes heart rate and blood sugar. A strength training expert agent processing module preprocesses the input data and inputs the preprocessed input data into the strength training expert agent, so that the strength training expert agent outputs the trainee's evaluation results based on prompt words and a strength training expert knowledge base. An evaluation result feedback module displays the evaluation results and determines the trainee's training plan based on the evaluation results. This approach overcomes the limitations of single-dimensional cognition through multimodal deep data acquisition. It relies on expert agents to achieve intelligent and precise reasoning feedback, replacing experience and simple rules. A continuous closed-loop optimization mechanism is established to promote dynamic iteration of training results. Compared with existing technologies, this system comprehensively upgrades the strength training auxiliary logic from data foundation, intelligent analysis, to training closed loop, providing more systematic technical support for scientific training and efficient improvement, adapting to the modern strength training requirements of precision, personalization, and long-term effectiveness.

[0063] The above-mentioned input data acquisition module is used to capture visual information through a camera; obtain muscle nerve signals through an electromyography sensor; and obtain physiological data through a detection bracelet.

[0064] The above-mentioned evaluation result feedback module is used to determine whether the evaluation result meets the preset training requirements; if so, generate maintenance suggestions and a training plan for the next stage; if not, generate correction suggestions and a training plan for the unmet training requirements.

[0065] The above-mentioned strength training action intelligent evaluation system based on multimodal fusion also includes: a strength training expert knowledge base construction module, which is used to obtain historical input data, and perform data cleaning and preprocessing on the historical input data; input the historical input data after data cleaning and preprocessing into the multimodal large model for fine-tuning training; construct an expert knowledge base based on the historical input data after data cleaning and preprocessing; and construct a strength training expert knowledge base based on the multimodal large model after fine-tuning training and the constructed expert knowledge base in combination with prompt words.

[0066] The above-mentioned strength training expert knowledge base construction module is also used to recognize action postures of historical visual information, and perform manual correction and data labeling; extract characteristic values ​​corresponding to target actions based on historical muscle nerve signals; and manually label historical physiological data of key nodes; among which, key nodes include: when the target muscle begins to exert force, when the target muscle is exhausted, and when the target muscle group is resting.

[0067] The above-mentioned strength training expert knowledge base construction module is used to input the characteristic values ​​of historical visual information and historical muscle nerve signals into the multimodal large model for fine-tuning training.

[0068] The above-mentioned strength training expert knowledge base construction module is used to construct an expert knowledge base based on historical physiological data of key nodes.

[0069] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the strength training action intelligent evaluation system based on multimodal fusion described above can refer to the corresponding process in the aforementioned embodiment and will not be repeated here.

[0070] Embodiment 5: The embodiment of the present invention also provides an electronic device for running the above-mentioned strength training action evaluation method; see Figure 8 The structure diagram of an electronic device shown in FIG. 1 includes a memory 100 and a processor 101, wherein the memory 100 is used to store one or more computer instructions, and the one or more computer instructions are executed by the processor 101 to implement the above-mentioned strength training action evaluation method.

[0071] Further, Figure 8 The electronic device shown further includes a bus 102 and a communication interface 103 , and the processor 101 , the communication interface 103 and the memory 100 are connected via the bus 102 .

[0072] The memory 100 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage. The communication connection between the system network element and at least one other network element is achieved through at least one communication interface 103 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used. The bus 102 may be an ISA bus, a PCI bus, or an EISA bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0073] The processor 101 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits in the processor 101 or software instructions. The above processor 101 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of the present invention can be directly embodied as being executed by a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium well-known in the art, such as a random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or register. The storage medium is located in the memory 100, and the processor 101 reads the information in the memory 100 and, in conjunction with its hardware, completes the steps of the method of the aforementioned embodiment.

[0074] An embodiment of the present invention also provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the above-mentioned strength training action evaluation method. The specific implementation can be found in the method embodiment, which will not be repeated here.

[0075] An embodiment of the present invention provides a computer program product for a strength training action intelligent evaluation system based on multimodal fusion, including a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the method in the previous method embodiment. For specific implementation, please refer to the method embodiment and will not be repeated here.

[0076] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working process of the system and / or device described above can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0077] In addition, in the description of the embodiments of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; they may refer to mechanical connections or electrical connections; they may refer to direct connections or indirect connections through an intermediate medium; and they may refer to internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0078] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0079] In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0080] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-described embodiments within the technical scope disclosed by the present invention, or replace some of the technical features therein with equivalents. Such modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A strength training action intelligent evaluation system based on multimodal fusion, characterized by: A method for evaluating a strength training action is used to perform the strength training action evaluation method, the strength training action evaluation method comprising: Collecting multimodal input data during a trainee's strength training; wherein the input data includes: visual information, muscle nerve signals, and physiological data of the trainee, the visual information includes: limb movements and posture trajectories, and the physiological data includes: heart rate and blood sugar; Preprocessing the input data, and inputting the preprocessed input data into a strength training expert agent, so that the strength training expert agent outputs an evaluation result of the trainee in combination with prompt words and a strength training expert knowledge base; The evaluation result is displayed, and a training plan for the trainee is determined based on the evaluation result.

2. The strength training action intelligent evaluation system based on multimodal fusion according to claim 1 is characterized in that: The steps for collecting multimodal input data include: capturing the visual information through a camera; Acquiring the muscle nerve signal through an electromyographic sensor; The physiological data is obtained by detecting the wristband.

3. The strength training action intelligent evaluation system based on multimodal fusion according to claim 1 is characterized in that: The step of determining the training plan of the trainee based on the evaluation result includes: Determining whether the evaluation result meets the preset training requirements; If yes, generate maintenance recommendations and a training plan for the next phase; If not, corrective suggestions and a training plan are generated for the training requirements that were not met.

4. The strength training action intelligent evaluation system based on multimodal fusion according to any one of claims 1 to 3, characterized in that: The method further comprises: Acquire historical input data, and perform data cleaning and preprocessing on the historical input data; Inputting the historical input data after data cleaning and preprocessing into the multimodal large model for fine-tuning training; Building the expert knowledge base based on the historical input data after data cleaning and preprocessing; In combination with the prompt words, the strength training expert knowledge base is constructed based on the multimodal large model after fine-tuning training and the constructed expert knowledge base.

5. The strength training action intelligent evaluation system based on multimodal fusion according to claim 4 is characterized in that: After the steps of cleaning and preprocessing the historical input data, the method further comprises: Perform action posture recognition on historical visual information, and perform manual correction and data annotation; Extract the feature value corresponding to the target action based on the historical muscle nerve signal; The historical physiological data of key nodes are manually annotated; wherein the key nodes include: when the target muscle starts to exert force, when the target muscle is exhausted, and when the target muscle group is resting.

6. The strength training action intelligent evaluation system based on multimodal fusion according to claim 5 is characterized in that: The step of inputting the historical input data after data cleaning and preprocessing into the multimodal large model for fine-tuning training includes: The characteristic values ​​of the historical visual information and the historical muscle nerve signals are input into a multimodal large model for fine-tuning training.

7. The strength training action intelligent evaluation system based on multimodal fusion according to claim 5 is characterized in that: The step of constructing the expert knowledge base based on the historical input data after data cleaning and preprocessing includes: The expert knowledge base is constructed based on the historical physiological data of key nodes.

8. A strength training action intelligent evaluation system based on multimodal fusion, characterized by: The strength training action intelligent evaluation system based on multimodal fusion includes: An input data acquisition module is used to collect multimodal input data during a trainee's strength training; wherein the input data includes visual information, muscle nerve signals, and physiological data of the trainee, wherein the visual information includes limb movements and posture trajectories, and the physiological data includes heart rate and blood sugar; A strength training expert agent processing module is used to pre-process the input data and input the pre-processed input data into the strength training expert agent, so that the strength training expert agent outputs the evaluation result of the trainee in combination with the prompt words and the strength training expert knowledge base; The evaluation result feedback module is used to display the evaluation result and determine the training plan of the trainee based on the evaluation result.

9. The strength training action intelligent evaluation system based on multimodal fusion according to claim 8 is characterized in that: The multimodal fusion-based strength training action intelligent evaluation system further includes: The strength training expert knowledge base construction module is used to construct the expert knowledge base based on the historical input data after data cleaning and preprocessing; combined with the prompt words, the strength training expert knowledge base is constructed based on the multimodal large model after fine-tuning training and the constructed expert knowledge base.

10. The strength training action intelligent evaluation system based on multimodal fusion according to claim 9 is characterized in that: The strength training expert knowledge base construction module is also used to perform action posture recognition on historical visual information, and perform manual correction and data annotation; extract the feature value corresponding to the target action based on the historical muscle nerve signal; The historical physiological data of key nodes are manually annotated; wherein the key nodes include: when the target muscle starts to exert force, when the target muscle is exhausted, and when the target muscle group is resting.

Images