Non-contact dynamic and static action training assistance method, device, equipment and medium

Abstract

The present application relates to the technical field of machine learning, and discloses a non-contact dynamic and static action training auxiliary method, device, equipment and medium, the method comprising: establishing a dynamic and static joint point inference model based on historical action training corresponding action image data combined with a preset deep learning model; real-time acquisition of action image data of a trainer training according to a preset action; obtaining the human body joint point data of the trainer through a preset recognition algorithm according to the action image data; obtaining dynamic and static feature data based on the dynamic and static joint point inference model, and performing data processing and fusion calculation on the human body joint point data to obtain real-time action index data; comparing the real-time action index data with preset action index data corresponding to the preset action, and assisting in guiding and prompting the action training corresponding to the action of the trainer according to the comparison result. Thus, the trainer can be better helped to perform action training, thereby improving training efficiency and safety.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, specifically to non-contact methods, devices, equipment, and storage media for assisting in training dynamic and static movements. Background Technology

[0002] Movement training is a common physical training method aimed at enhancing muscle strength, endurance, and explosive power. Dynamic movement training primarily involves weight training using external loads (such as dumbbells, barbells, etc.) to activate and exercise muscles through various movements. Static movement training mainly enhances muscle strength by maintaining posture or continuous muscle contraction. However, traditional movement training methods often lack real-time feedback mechanisms due to limitations such as time and cost. Not every trainee can receive professional guidance and supervision, making it difficult for them to immediately understand their training results and the correctness of their movements. This makes it hard to adjust posture and technique, potentially leading to the continued use of incorrect training postures and techniques, increasing the risk of injury, and compromising the trainee's safety. Summary of the Invention

[0003] In view of this, the present invention provides a non-contact dynamic and static motion training assistance method to solve the problem of low accuracy and safety of training movements during the training process.

[0004] In a first aspect, the present invention provides a non-contact method for assisting in dynamic and static motion training. The method includes: establishing a dynamic and static joint point inference model based on motion image data corresponding to historical motion training and a preset deep learning model; acquiring motion image data of a trainee training according to preset motions in real time, wherein the image data is obtained through an image acquisition device set at a preset distance from the trainee; obtaining human joint point data of the trainee through a preset recognition algorithm based on the motion image data, and simultaneously obtaining dynamic and static feature data based on the dynamic and static joint point inference model; performing data processing and fusion calculation on the dynamic and static feature data and the human joint point data to obtain real-time motion index data; comparing the real-time motion index data with the preset motion index data corresponding to the preset motions, and providing auxiliary guidance prompts for the trainee's motion training corresponding to the motions using a display device based on the comparison results.

[0005] This invention establishes a dynamic and static joint inference model by combining historical motion image data with a preset deep learning model. It collects image data in real time and uses computer vision technology to analyze the trainee's movements to obtain real-time motion indicators. These real-time indicators are then compared with preset motion indicators to provide auxiliary guidance. By combining image processing, deep learning, and data fusion technologies, it achieves real-time monitoring and auxiliary guidance of the trainee's motion training process, thereby helping trainees to better perform motion training and improving training efficiency and safety.

[0006] In one optional implementation, the step of establishing a dynamic and static joint inference model based on motion image data corresponding to historical motion training and a preset deep learning model includes: extracting data from motion image data corresponding to multiple historical motion trainings to obtain dynamic joint data and static joint data; calibrating the dynamic joint data and static joint data to obtain a training set; and training a preset machine learning classification task model based on the training set to establish a dynamic and static joint inference model.

[0007] By converting historical motion image data corresponding to motion training into a training set, a dynamic and static joint inference model is established after training. This inference model accurately infers the joint positions and dynamic and static feature data of the motions corresponding to the motion training, improving the recognition accuracy of the motions and providing a reliable foundation for subsequent data processing and fusion calculations.

[0008] In one optional implementation, the step of training a preset machine learning classification task model based on the training set to establish a dynamic and static joint inference model includes: performing feature engineering on the training set; inputting the feature-engineered training set into a preset deep learning model for training; evaluating the performance of the preset deep learning model using a preset test set; and using the model that meets the preset performance evaluation standard as the dynamic and static joint inference model.

[0009] By performing feature engineering on the training set, effective features can be extracted, key features can be highlighted, input data can be optimized, the accuracy of training data and the learning ability of the model can be improved, the model performance can be evaluated, and the trained inference model can be guaranteed to have high accuracy.

[0010] In one optional implementation, the step of obtaining the trainee's human joint data from the motion image data using a preset recognition algorithm, and simultaneously obtaining dynamic and static feature data based on the dynamic and static joint inference model, includes: performing noise reduction processing on the motion image data to obtain processed image data; using the preset recognition algorithm to perform human detection and localization on the processed image data to obtain the trainee's human joint data; and inferring dynamic joint feature data and static joint feature data through the dynamic and static joint inference model.

[0011] By denoising motion image data, the quality of the image data is improved and the interference of noise in the image is reduced. Using a preset recognition algorithm, the processed image data is used to detect and locate human bodies. The position and posture of human bodies in the image can be accurately found. Dynamic and static feature data are obtained through model inference, thereby more comprehensively representing all aspects of the training motion. This provides a foundation for subsequent data processing and fusion calculations combined with the inference model.

[0012] In one optional implementation, the step of processing and fusing the dynamic and static feature data and the human joint point data to obtain real-time motion index data includes: processing and fusing the human joint point data, dynamic joint point feature data and static joint point feature data to obtain real-time motion index data.

[0013] By processing and fusing human joint point data, dynamic joint point feature data, and static joint point feature data, real-time motion index data can be obtained. This allows for a quantitative assessment of the trainee's performance and effectiveness when performing the corresponding motion training movements. The results can be compared with preset motion indexes to provide a basis for evaluation for the trainee.

[0014] In one optional implementation, the step of comparing the real-time motion index data with the preset motion index data corresponding to the preset motion, and providing auxiliary guidance prompts to the trainee's motion training using a display device based on the comparison result, includes: comparing the real-time motion index data with the preset motion index data; if the comparison result is the same, providing correct voice prompts to the trainee; if the comparison result is different, determining the trainee's incorrect motion information based on the real-time motion index data, providing incorrect voice prompts to the trainee based on the incorrect motion information, and providing corrective feedback prompts using a display device.

[0015] By comparing real-time motion indicator data with preset motion indicator data, the system provides auxiliary guidance and prompts for trainees to perform corresponding movements based on the comparison results. Correct voice prompts enhance trainees' confidence and accuracy in executing movements, while incorrect movement information is accompanied by incorrect voice prompts and corrective feedback displayed on the screen, helping trainees recognize errors and thus improving the accuracy and effectiveness of their movements.

[0016] In one optional implementation, after comparing the real-time motion index data with the preset motion index data corresponding to the preset motion, and providing auxiliary guidance prompts to the trainee's motion training using a display device based on the comparison results, the method further includes: continuously monitoring and recording the training data during the trainee's training process; and generating a training report based on the training data.

[0017] By continuously monitoring and recording training data during the training process, comprehensive training information can be provided, generating training reports that allow for a comprehensive analysis of the trainee's performance. Personalized reports are offered to each trainee, facilitating the development of subsequent training plans and ultimately leading to more effective movement training.

[0018] Secondly, the present invention provides a non-contact dynamic and static motion training aid device, the device comprising:

[0019] The model building module is used to build a dynamic and static joint inference model based on the motion image data corresponding to historical motion training and a preset deep learning model.

[0020] The data acquisition module is used to acquire motion image data of the trainee training according to preset movements in real time. The image data is obtained through an image acquisition device set at a preset distance from the trainee.

[0021] The joint point and feature recognition module is used to obtain the trainee's human joint point data through a preset recognition algorithm based on the motion image data, and at the same time obtain dynamic and static feature data based on the dynamic and static joint point inference model.

[0022] The fusion computing module is used to process and fuse the dynamic and static feature data and the human joint point data to obtain real-time motion index data.

[0023] The action comparison and guidance module is used to compare the real-time action index data with the preset action index data corresponding to the preset action, and to provide auxiliary guidance and prompts for the trainee's action training based on the comparison results using a display device.

[0024] Thirdly, the present invention provides a computer device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the non-contact dynamic and static motion training assistance method described in the first aspect or any corresponding embodiment thereof.

[0025] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the non-contact dynamic and static motion training assistance method described in the first aspect or any corresponding embodiment thereof. Attached Figure Description

[0026] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0027] Figure 1 This is a flowchart illustrating a non-contact dynamic and static motion training assistance method according to an embodiment of the present invention.

[0028] Figure 2 This is a flowchart illustrating another non-contact dynamic and static motion training assistance method according to an embodiment of the present invention;

[0029] Figure 3 This is a flowchart illustrating another non-contact dynamic and static motion training assistance method according to an embodiment of the present invention;

[0030] Figure 4 This is a flowchart illustrating another non-contact dynamic and static motion training assistance method according to an embodiment of the present invention;

[0031] Figure 5 This is an activity diagram of another non-contact dynamic and static motion training assistance method according to an embodiment of the present invention;

[0032] Figure 6 This is a schematic diagram of the training action interface of another non-contact dynamic and static motion training assistance method according to an embodiment of the present invention.

[0033] Figure 7 This is a schematic diagram of the training action interface of another non-contact dynamic and static motion training assistance method according to an embodiment of the present invention.

[0034] Figure 8 This is a schematic diagram of the module composition of a non-contact dynamic and static motion training auxiliary device according to an embodiment of the present invention;

[0035] Figure 9 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] This invention provides a non-contact dynamic and static motion training assistance method, applied in scenarios where trainees perform motion training. It utilizes a non-contact dynamic and static motion training assistance device to capture images of the trainee's movements in real time, identify human joint points, and perform data processing and fusion calculations using a dynamic and static reasoning model. This helps trainees intuitively see the implementation of their movements and provides auxiliary guidance such as voice and text, displayed on a display device. This prevents trainees from making incorrect movements that lead to substandard training results and provides correct prompts for correct movements, thereby helping trainees to better perform motion training and improving training efficiency and safety.

[0038] According to an embodiment of the present invention, a non-contact dynamic and static motion training assistance method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0039] This embodiment provides a non-contact method for assisting in dynamic and static motion training, which can be used with the aforementioned computer equipment. Figure 1 This is a flowchart of a non-contact dynamic and static motion training assistance method according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps:

[0040] Step S101: Based on the motion image data corresponding to the historical motion training, a dynamic and static joint inference model is established by combining the preset deep learning model.

[0041] In this embodiment of the invention, the motion image data corresponding to historical motion training refers to image data obtained by historically photographing or recording the motion process of a trainee during motion training, or image data obtained based on a motion training database. Image data can include the trainee's posture, motion path, and joint position at different points in time. The preset deep learning model refers to a pre-designed and trained deep learning model, used in this embodiment to train the dynamic and static joint inference model. The dynamic and static joint inference model combines image data with the preset deep learning model to infer dynamic and static feature data in the image. Dynamic and static feature data includes static joint feature data and dynamic joint feature data. Static joint feature data includes angle information, length information, posture, etc., while dynamic joint feature data includes motion speed, motion trajectory, acceleration, etc.

[0042] It should be noted that dynamic characteristics refer to the movement of the body when multiple joints are connected during exercise training, and this phenomenon is described as dynamic characteristics; static characteristics refer to the relatively still body when multiple joints are connected during exercise training, and this phenomenon is described as static characteristics.

[0043] In one specific embodiment, a set of image data corresponding to historical action training is provided. For example, this image data records the process of a trainee performing squat training. A preset deep learning model, such as an action recognition model based on a convolutional neural network, is then used to train this image data to establish a dynamic and static joint inference model. The model can infer the dynamic and static feature data of the action from the images.

[0044] Step S102: Real-time acquisition of motion image data of the trainee training according to preset movements.

[0045] In this embodiment of the invention, it is understood that the image data is obtained through an image acquisition device set at a preset distance from the trainee (as long as a clear and complete training image of the trainee can be obtained). The motion image data is acquired in real time, which is convenient for subsequent comparison with the trainee's real-time motion index data and preset motion index data to evaluate the trainee's motion accuracy, whether the posture is correct, and whether the range of motion is sufficient. At the same time, it also avoids safety issues that may occur when the trainee trains alone without guidance.

[0046] In one specific embodiment, an image acquisition device (e.g., a camera) positioned at a preset distance from the trainee captures the trainee's posture and movement process in real time during the corresponding exercise. Taking squats as an example, as the trainee begins training according to a preset squatting motion, the camera records multiple keyframes of image or video data as motion image data during the process from standing up to squatting down and returning to standing up.

[0047] Step S103: Based on the motion image data, obtain the trainee's human joint data through a preset recognition algorithm, and at the same time obtain dynamic and static feature data based on the dynamic and static joint inference model.

[0048] In this embodiment of the invention, the preset recognition algorithm refers to a pre-trained algorithm model used to analyze and process image data to extract key information. In this embodiment, the preset recognition algorithm can be used to identify key human joints from image data of actions corresponding to motion training. Human joint data is obtained by performing joint inference or tracking on the human body to determine the positional information of each joint.

[0049] In one optional embodiment, the present invention can also acquire human joint point data through various non-contact devices, including but not limited to: imaging devices, infrared devices, radar devices, laser devices, ultrasound devices, etc., and the embodiments of the present invention do not limit this.

[0050] Step S104: Perform data processing and fusion calculation on the dynamic and static feature data and the human joint data to obtain real-time motion index data.

[0051] In this embodiment of the invention, after processing and fusing human joint point data and dynamic / static feature data, real-time motion index data can be obtained. Real-time motion index data can reflect the execution status of the corresponding movement during training, such as the accuracy of joint point angles. For example, when a trainee is performing a squat, the joint features of the trainee's knee, thigh, and calf are obtained through motion image data and the dynamic / static joint point inference model. Combined with joint point position data, the trainee's knee flexion angle is calculated to be 60 degrees, which is the real-time motion index data.

[0052] Step S105: Compare the real-time motion index data with the preset motion index data corresponding to the preset motion, and provide auxiliary guidance prompts for the trainee's motion training using a display device based on the comparison results.

[0053] In this embodiment of the invention, the preset movement index data are standard parameter indicators set for each specific movement training movement, representing a range of preset indicators. These standard parameter indicators include static parameter indicators and dynamic parameter indicators. Static parameter indicators include, but are not limited to, the angles, amplitudes, and durations of static joints throughout the body. Dynamic parameter indicators include, but are not limited to, the angles, speeds, and amplitudes of dynamic joints throughout the body. This embodiment of the invention does not impose any limitations on these. It should be noted that the dynamic parameter indicators also include static jitter data. Static jitter data can be understood as the slight tremors that occur when the human body needs to maintain a movement for a long time, as it cannot remain completely still. Therefore, static jitter data is also included as a dynamic parameter indicator in the evaluation range of the preset movement index. Auxiliary guidance prompts refer to providing real-time guidance and feedback to the trainee based on the comparison results between real-time movement index data and preset movement index data. By analyzing the differences between the two index data, problems in the trainee's movement execution are identified, and prompts or guidance are provided to the trainee to help improve their posture.

[0054] In one optional implementation, after providing auxiliary guidance and prompts to the trainee, all the collected data obtained in the above steps will be stored in the training database to facilitate subsequent demonstrations for other trainees or the current trainee, or as training data for model training based on historical training actions.

[0055] This invention establishes a dynamic and static joint inference model by combining historical motion image data with a preset deep learning model. It collects image data in real time and uses computer vision technology to analyze the trainee's movements to obtain real-time motion indicators. These real-time indicators are then compared with preset motion indicators to provide auxiliary guidance. This invention combines image processing, deep learning, and data fusion technologies to achieve real-time monitoring and auxiliary guidance of the trainee's motion training process, thereby helping trainees to better perform motion training and improving training efficiency and safety.

[0056] This embodiment provides a non-contact method for assisting in the training of dynamic and static movements, which can be used in the aforementioned computers, etc. Figure 2 This is a flowchart of a non-contact dynamic and static motion training assistance method according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps:

[0057] Step S201 involves establishing a dynamic and static joint inference model based on the motion image data corresponding to historical motion training and a pre-set deep learning model. Specifically, step S201 includes:

[0058] Step S2011: Extract motion image data corresponding to multiple historical motion trainings to obtain dynamic joint data and static joint data.

[0059] For example, if training is performed on a model of a trainee performing a squatting motion, dynamic data (joint points) of the hip, knee, and ankle joints related to the motion, as well as static data of the body (such as the amplitude of the upper body relative to the ground) are extracted from multiple historical image data.

[0060] Step S2012: Perform data calibration on the dynamic joint data and the static joint data to obtain a training set.

[0061] In practice, dynamic and static joints are calibrated and matched with corresponding action poses to form a calibration dataset for training. The training set contains dynamic and static joint data related to each action.

[0062] Step S2013: Based on the training set, train the preset machine learning classification task model to establish a dynamic and static key point reasoning model.

[0063] For example, a dynamic and static joint inference model for squatting is trained using a convolutional neural network (CNN) combined with a recurrent neural network (RNN) structure, taking dynamic and static joint data of the squatting motion as input. This model infers the accuracy of the squatting motion. Understandably, through continuous training and learning, the dynamic and static joint inference model can be continuously optimized to adapt to inference for multiple training exercises.

[0064] Specifically, step S2013 includes:

[0065] a1: Perform feature engineering on the training set.

[0066] It is understandable that redundant and irrelevant features may exist in both dynamic and static keypoint data. Through feature engineering, the features needed for model training can be selected to improve the model's inference ability.

[0067] a2: Input the feature-engineered training set into the preset deep learning model for training.

[0068] Understandably, by iteratively training the training set, the model can be adjusted based on the input data and labels to minimize the difference between the prediction and the true label.

[0069] a3: The performance of the preset deep learning model is evaluated using a preset test set, and the model that meets the preset performance evaluation standard is used as the dynamic and static joint inference model.

[0070] Understandably, the test set consists of data samples different from the training set, used to evaluate the generalization ability of a trained model on unseen data. In model performance evaluation, metrics include accuracy, precision, and recall. In practice, samples from the test set are input into the model one by one. The model infers for each sample, obtaining a predicted label ("correct" or "incorrect"). These correct or incorrect labels are then classified, and evaluation metrics are calculated using the classified samples to further optimize the model or adjust its parameters.

[0071] In this embodiment, the motion image data corresponding to historical motion training is converted into a training set, and a dynamic and static joint inference model is established after training. The inference model accurately infers the joint positions and dynamic features in the motions corresponding to the motion training, improving the recognition accuracy of the motions and providing a reliable foundation for subsequent data processing and fusion calculations. Simultaneously, feature engineering processing of the training set extracts effective features, highlights key features, optimizes the input data, improves the accuracy of the training data and the model's learning ability, and evaluates the model's performance to ensure that the trained inference model has high accuracy.

[0072] Step S202: Acquire motion image data of the trainee performing training according to preset movements in real time. For details, please refer to [link to relevant documentation]. Figure 1 Step S102 of the illustrated embodiment will not be described again here.

[0073] Step S203: Based on the motion image data, obtain the trainee's human joint point data through a preset recognition algorithm, and simultaneously obtain dynamic and static feature data based on the dynamic and static joint point inference model. For details, please refer to [link to details]. Figure 1 Step S103 of the illustrated embodiment will not be described again here.

[0074] Step S204 involves processing and fusing the dynamic and static feature data and the human joint point data to obtain real-time motion index data. For details, please refer to [link to relevant documentation]. Figure 1 Step S104 of the illustrated embodiment will not be described again here.

[0075] Step S205: Compare the real-time motion index data with the preset motion index data corresponding to the preset motion, and provide auxiliary guidance and prompts for the trainee's motion training using a display device based on the comparison result. For details, please refer to [link to details]. Figure 1 Step S105 of the illustrated embodiment will not be described again here.

[0076] This embodiment provides a non-contact method for assisting in the training of dynamic and static movements, which can be used in the aforementioned computers, etc. Figure 3 This is a flowchart of a non-contact dynamic and static motion training assistance method according to an embodiment of the present invention, such as... Figure 3 As shown, the process includes the following steps:

[0077] Step S301: Based on the motion image data corresponding to historical motion training, a dynamic and static joint inference model is established using a pre-set deep learning model. For details, please refer to [link to details]. Figure 1 Step S101 of the illustrated embodiment will not be described again here.

[0078] Step S302: Acquire motion image data of the trainee performing training according to preset movements in real time. For details, please refer to [link to relevant documentation]. Figure 1 Step S102 of the illustrated embodiment will not be described again here.

[0079] Step S303 involves obtaining the trainee's human joint point data from the motion image data using a preset recognition algorithm, and simultaneously obtaining dynamic and static feature data based on the dynamic and static joint point inference model. Specifically, step S303 includes:

[0080] Step S3031: Denoise reduction processing is performed on the motion image data to obtain processed image data.

[0081] For example, the motion images corresponding to the motion training can be smoothed by using image filtering methods, such as Gaussian filters, to reduce the impact of noise by adjusting the filter parameters, such as filter size and standard deviation.

[0082] Step S3032: Use a preset recognition algorithm to perform human body detection and localization on the processed image data to obtain the trainee's human joint data.

[0083] For example, human detection and localization are performed on the processed image data. The human body region in the image is located and extracted using a recognition algorithm to obtain localization information, such as the coordinates of multiple joints of the human body.

[0084] Step S3033: The motion image data is inferred through the dynamic and static joint inference model to obtain dynamic joint feature data and static joint feature data.

[0085] For example, the dynamic and static joint inference model performs inference to obtain the trainee's joint movement speed (dynamic feature) and angle (static feature).

[0086] It should be noted that the above steps S3031, S3032 and S3033 are not in a sequential order, but are only illustrative examples of the images.

[0087] In this embodiment, noise reduction processing is performed on the motion image data to improve the quality of the image data and reduce the interference of noise in the image. The human body detection and positioning are performed on the processed image data using a preset recognition algorithm, which can accurately find the position and posture of the human body in the image. Dynamic and static feature data are obtained through model inference, thereby more comprehensively representing all aspects of the training action and providing a foundation for subsequent data processing and fusion calculation in combination with the inference model.

[0088] Step S304 involves processing and fusing the dynamic and static feature data and the human joint point data to obtain real-time motion index data. Specifically, step S304 includes:

[0089] Step S3041: Real-time motion index data is obtained by processing and fusing the human joint point data, dynamic joint point feature data, and static joint point feature data.

[0090] For example, based on the coordinate information of multiple joints in the human body, combined with the movement speed (dynamic feature) and angle (static feature), the joint trajectory and joint angular velocity of the movement corresponding to the movement training are calculated as real-time movement indicators.

[0091] In this embodiment, by processing and fusing human joint data, dynamic joint feature data, and static joint feature data, real-time motion index data can be obtained, thereby quantitatively evaluating the trainee's performance and effect when performing the corresponding motion training. This data is then compared with preset motion indicators to provide an evaluation basis for the trainee.

[0092] Step S305: Compare the real-time motion index data with the preset motion index data corresponding to the preset motion, and provide auxiliary guidance and prompts for the trainee's motion training using a display device based on the comparison result. For details, please refer to [link to details]. Figure 1 Step S105 of the illustrated embodiment will not be described again here.

[0093] This embodiment provides a non-contact method for assisting in the training of dynamic and static movements, which can be used in the aforementioned computers, etc. Figure 4 This is a flowchart of a non-contact dynamic and static motion training assistance method according to an embodiment of the present invention, such as... Figure 4 As shown, the process includes the following steps:

[0094] Step S401: Based on the motion image data corresponding to historical motion training, a dynamic and static joint inference model is established using a pre-set deep learning model. For details, please refer to [link to details]. Figure 1 Step S101 of the illustrated embodiment will not be described again here.

[0095] Step S402: Acquire motion image data of the trainee performing training according to preset movements in real time. For details, please refer to [link to relevant documentation]. Figure 1 Step S102 of the illustrated embodiment will not be described again here.

[0096] Step S403: Based on the motion image data, obtain the trainee's human joint point data through a preset recognition algorithm, and simultaneously obtain dynamic and static feature data based on the dynamic and static joint point inference model. For details, please refer to [link to details]. Figure 1 Step S103 of the illustrated embodiment will not be described again here.

[0097] Step S404 involves processing and fusing the dynamic and static feature data and the human joint point data to obtain real-time motion index data. For details, please refer to [link to relevant documentation]. Figure 1 Step S104 of the illustrated embodiment will not be described again here.

[0098] Step S405: Compare the real-time motion index data with the preset motion index data corresponding to the preset motion, and use the display device to provide auxiliary guidance and prompts for the corresponding motion training of the trainee based on the comparison results.

[0099] Specifically, step S405 above includes:

[0100] Step S4051: Compare the real-time motion indicator data with the preset motion indicator data.

[0101] For example, preset movement (e.g., squat) metrics require the trainee to achieve a preset knee flexion angle and a preset upper limb vertical movement speed. This involves comparing the knee flexion angle and upper limb vertical movement speed in real-time movement metrics with the corresponding values ​​in preset movement metrics.

[0102] In step S4052, if the comparison results are the same, the trainee is given the correct voice prompt; if the comparison results are different, the trainee's incorrect movement information is determined based on the real-time movement index data, and the trainee is given incorrect voice prompts and corrective feedback prompts based on the incorrect movement information.

[0103] For example, if the comparison results are the same, meaning the real-time motion indicator data matches the preset motion indicator data, it indicates that the trainee's movement execution is correct. Positive encouragement is given through voice prompts, such as "Movement correct, please continue," or text prompts. If the comparison results are different, meaning the real-time motion indicator data does not match the preset motion indicator data, it indicates that the trainee's movement execution has errors. Based on the real-time motion indicator data, the specific information of the error is determined, such as the knee flexion angle being too large or too small, or the upper limb's vertical movement speed being too slow or too fast. Voice or text prompts, such as "Movement angle too large, please adjust appropriately" or "Movement speed too slow, please speed up," will be displayed on the device to help the trainee correct the errors.

[0104] In this embodiment, real-time motion index data is compared with preset motion index data, and auxiliary guidance prompts are provided to the trainee for the corresponding motion training based on the comparison results. Correct voice prompts enhance the trainee's confidence and the accuracy of motion execution, while incorrect motion information is accompanied by incorrect voice prompts and corrective feedback displayed on the screen, helping the trainee to recognize errors and thereby improving the accuracy and effectiveness of the trainee's movements.

[0105] Step S406: Continuously monitor and record the training data of the trainee during the training process.

[0106] In this embodiment, training data refers to relevant data generated by the trainee during the movement training process, such as the execution status of the movements. Through a camera device positioned within a preset range of the trainee, the training data is recorded and monitored in real time. Throughout the entire training process, the completion rate of each movement by the trainee, as well as the auxiliary guidance prompts, are monitored.

[0107] Step S407: Generate a training report based on the training data.

[0108] In this embodiment, the training report is generated based on training data and is used to evaluate the trainee's training process and training results. Based on the training data, a training report can be generated, which may include information such as the trainee's performance on movements. For example, for a squat exercise, based on the assisted guidance prompts, the number of incorrect and correct repetitions can be calculated, resulting in a performance rate of 80%.

[0109] Reference Figure 5 , Figure 5 This is an activity diagram for motion training in this invention. The interactive page acquires the motion selected by the trainee and the corresponding motion for training. At this time, the dynamic and static model algorithms are loaded, and the image acquisition device is also activated. The motion training interface is as follows: Figure 6As shown, the system displays the current training status to the trainee. Taking single-leg standing as an example, when the trainee makes an incorrect movement, the corresponding joint point is highlighted in red, and the incorrect part is marked with an arc. The display shows the real-time knee flexion angle, and real-time voice correction is provided. The image acquisition device monitors the interface status in real time. When exiting the interface, the trainee can reselect the corresponding movement for training. When the trainee remains on the interface, the real-time image data is used to obtain the coordinate data of all relevant joint positions on the whole body, and dynamic and static feature data are calculated through an inference model. After fusing the dynamic and static feature data and the coordinate data of the joint positions, the real-time movement index is calculated and compared with the selected movement index (preset movement index) to determine the trainee's training status and provide auxiliary guidance, which is also displayed on the page. Data is continuously recorded during training, and a training report is generated after the trainee finishes training to help the trainee evaluate the training effect.

[0110] Reference Figure 7 , Figure 7 This is a diagram illustrating the forward knee extension and straight leg raise training exercise of this invention. The trainee needs to raise the hip, knee, and ankle of the training side forward while maintaining the angle of the three joints at approximately 180 degrees, and keeping the upper body upright, until the preset angle is reached. Figure 7 To achieve a 45-degree angle between the legs, during training, the first straight line connecting the ankle (a), knee (b), and hip (c) joints remains relatively stationary. Therefore, the indicators to be compared on this first straight line are the static parameters in the preset action indicator data, such as the knee angle (corresponding to static features). However, the first straight line connecting the ankle (a), knee (b), and hip (c) joints remains in motion relative to the body. Therefore, during training, the indicators to be compared on the body perpendicular to the ground are the dynamic parameters in the preset action indicator data, such as the leg lifting angle and speed (corresponding to dynamic features).

[0111] In this embodiment, continuous monitoring and recording of training data during the training process can provide comprehensive training information, generate training reports, and comprehensively analyze the trainee's performance. Personalized reports are provided to each trainee, facilitating the development of subsequent training plans and ultimately achieving more effective movement training.

[0112] This embodiment also provides a non-contact dynamic and static motion training aid device, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0113] This embodiment provides a non-contact dynamic and static motion training aid device, such as... Figure 8 As shown, it includes:

[0114] The model building module 801 is used to build a dynamic and static joint inference model based on the action image data corresponding to historical action training and a preset deep learning model.

[0115] The data acquisition module 802 is used to acquire motion image data of the trainee training according to preset movements in real time. The image data is obtained through an image acquisition device set at a preset distance from the trainee.

[0116] The joint point and feature recognition module 803 is used to obtain the human joint point data of the trainee through a preset recognition algorithm based on the motion image data, and at the same time obtain dynamic and static feature data based on the dynamic and static joint point inference model.

[0117] The fusion computing module 804 is used to process and fuse the dynamic and static feature data and the human joint data to obtain real-time motion index data.

[0118] The action comparison and guidance module 805 is used to compare the real-time action index data with the preset action index data corresponding to the preset action, and to provide auxiliary guidance and prompts for the trainee's action training based on the comparison results using a display device.

[0119] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0120] In this embodiment, the non-contact dynamic and static motion training aid is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0121] This invention also provides a computer device having the above-described features. Figure 8 The non-contact dynamic and static motion training aid device shown is shown.

[0122] Please see Figure 9 , Figure 9 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 9As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 9 Take a processor 10 as an example.

[0123] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0124] The memory 20 stores instructions executable by at least one processor 10 to cause at least one processor 10 to perform the method shown in the above embodiments.

[0125] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0126] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0127] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0128] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0129] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and all such modifications and variations fall within the scope defined herein.

Claims

1. A non-contact method for assisting in dynamic and static motion training, characterized in that, The method includes: A dynamic and static joint inference model is established based on motion image data corresponding to historical motion training and a pre-set deep learning model, including: Data extraction is performed on motion image data corresponding to multiple historical motion training to obtain dynamic joint data and static joint data. The dynamic and static joint data are calibrated to obtain a training set. The preset machine learning classification task model is trained based on the training set to establish the dynamic and static key point reasoning model. The preset machine learning classification task model is a structural model that combines convolutional neural networks and recurrent neural networks. Real-time acquisition of motion image data of trainees training according to preset movements, wherein the image data is obtained through an image acquisition device set at a preset distance from the trainee; Based on the motion image data, the trainee's human joint data is obtained through a preset recognition algorithm, and dynamic and static feature data is obtained based on the dynamic and static joint inference model. The dynamic and static feature data includes dynamic joint feature data and static joint feature data. The dynamic joint feature data is used to characterize the movement of the body connected by multiple joints when the trainee performs movement training. The static joint feature data is used to characterize the body connected by multiple joints when the trainee performs movement training. The dynamic and static feature data and the human joint point data are processed and fused to obtain real-time motion index data, including: Real-time motion index data is obtained by processing and fusing the human joint point data, dynamic joint point feature data and static joint point feature data. The real-time motion index data is compared with the preset motion index data corresponding to the preset motion, and the corresponding motion training for the trainee is assisted and guided by the display device based on the comparison results. The preset motion index data includes static parameter indexes and dynamic parameter indexes. The static parameter indexes include the angle, amplitude, and duration of static joints throughout the body; the dynamic parameter indexes include the angle, speed, amplitude, and static shaking data of dynamic joints throughout the body.

2. The method according to claim 1, characterized in that, The step of training a preset machine learning classification task model based on the training set to establish a dynamic and static key point reasoning model includes: The training set is subjected to feature engineering processing; The feature-engineered training set is input into a pre-defined deep learning model for training; The performance of the preset deep learning model is evaluated using a preset test set, and the model that meets the preset performance evaluation standard is used as the dynamic and static joint inference model.

3. The method according to claim 1, characterized in that, The process of obtaining the trainee's human joint data from the motion image data using a preset recognition algorithm, and simultaneously obtaining dynamic and static feature data based on the dynamic and static joint inference model, includes: The motion image data is denoised to obtain the processed image data; The processed image data is used to perform human body detection and localization using a preset recognition algorithm to obtain the human body joint data of the trainee. Dynamic and static joint feature data are obtained through reasoning using the dynamic and static joint inference model.

4. The method according to any one of claims 1 to 3, characterized in that, The step of comparing the real-time motion index data with the preset motion index data corresponding to the preset motion, and providing auxiliary guidance and prompts for the trainee's motion training using a display device based on the comparison results, includes: The real-time motion indicator data is compared with the preset motion indicator data; If the comparison results are the same, then the trainee will receive the correct voice prompt. If the comparison results are different, the trainee's erroneous movement information is determined based on the real-time movement index data. The trainee is then given erroneous voice prompts based on the erroneous movement information, and corrective feedback is provided through the display device.

5. The method according to claim 4, characterized in that, After comparing the real-time motion index data with the preset motion index data corresponding to the preset motion, and providing auxiliary guidance and prompts for the trainee's motion training using a display device based on the comparison results, the method further includes: Continuously monitor and record training data during the trainee's training process; A training report is generated based on the training data.

6. A non-contact device for assisting in dynamic and static motion training, characterized in that, The device includes: The model building module is used to build a dynamic and static joint inference model based on motion image data corresponding to historical motion training and a preset deep learning model. Specifically, the model building module is used to extract data from motion image data corresponding to multiple historical motion trainings to obtain dynamic joint data and static joint data; to perform data labeling on the dynamic and static joint data to obtain a training set; and to train a preset machine learning classification task model based on the training set to build the dynamic and static joint inference model. The preset machine learning classification task model is a structure model combining convolutional neural networks and recurrent neural networks. The data acquisition module is used to acquire motion image data of the trainee training according to preset movements in real time. The image data is obtained through an image acquisition device set at a preset distance from the trainee. The joint point and feature recognition module is used to obtain the trainee's human joint point data through a preset recognition algorithm based on the motion image data, and at the same time obtain dynamic and static feature data based on the dynamic and static joint point inference model; wherein, the dynamic and static feature data includes dynamic joint point feature data and static joint point feature data, the dynamic joint point feature data is used to characterize the movement of the body connected by multiple joint points when the trainee performs motion training, and the static joint point feature data is used to characterize the body connected by multiple joint points when the trainee performs motion training. The fusion calculation module is used to process and fuse the dynamic and static feature data and the human joint point data to obtain real-time motion index data; specifically, the fusion calculation module is used to process and fuse the human joint point data, dynamic joint point feature data and static joint point feature data to obtain real-time motion index data. The motion comparison and guidance module is used to compare the real-time motion index data with the preset motion index data corresponding to the preset motion, and to provide auxiliary guidance prompts for the trainee's motion training using a display device based on the comparison results. The preset motion index data includes static parameter indicators and dynamic parameter indicators. The static parameter indicators include the angle, amplitude, and duration of static joints throughout the body; the dynamic parameter indicators include the angle, speed, amplitude, and static shaking data of dynamic joints throughout the body.

7. A computer device, characterized in that, include: A memory and a processor are communicatively connected, the memory stores computer instructions, and the processor executes the computer instructions to perform the non-contact dynamic and static motion training assistance method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to perform the non-contact dynamic and static motion training assistance method according to any one of claims 1 to 5.

Images