Multi-angle body posture analysis and evaluation method and device

Abstract

This invention provides a multi-angle posture analysis and assessment method and device, comprising: non-contact acquisition of human posture data from the front, side, back, and top of the human body using a wide-angle camera; image distortion correction and standardization of the acquired image data; extraction of posture abnormality assessment features, selection of key human body surface feature coordinates from multiple angles, and construction of a feature set; extraction of joint coordinates using an improved YOLO pose estimation algorithm and a UNet-V2 image segmentation algorithm, and training a posture assessment model; analysis of the subject's posture images using the posture assessment model, and output of a comprehensive score and evaluation report; periodic updating of model parameters based on accumulated subject data; and remote storage, management, and transmission of data via a network system.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of pattern recognition technology, and in particular to a method and device for multi-angle posture analysis and evaluation. Background Technology

[0002] With the rapid development of economy and technology, the public's pursuit of a scientific and healthy life is increasing, and postural health has received much attention because it can prevent joint and spinal diseases and enhance psychological self-confidence. However, increased study and work pressure, prolonged sitting, insufficient exercise, and excessive reliance on electronic devices have led to frequent postural abnormalities such as forward head posture, hunchback, and scoliosis. Postural examination is divided into static and dynamic examinations. Static examinations assess structural problems by observing body symmetry and using imaging equipment, while dynamic examinations identify functional compensation problems through motion testing. However, traditional methods suffer from drawbacks such as reliance on human assessment, high subjectivity, and low efficiency. Although deep learning-based pose estimation technology has been applied, it faces limitations such as high cost of detection equipment, poor portability, lack of scientific rigor in single-angle observation, and insufficient professionalism in evaluation indicators. Summary of the Invention

[0003] To address the aforementioned problems in the existing technology, embodiments of the present invention provide a multi-angle body posture analysis and evaluation method and device.

[0004] In a first aspect, embodiments of the present invention provide a multi-angle posture analysis and evaluation method, comprising: non-contactly acquiring posture data of the front, side, back, and top of a human body using a wide-angle camera; correcting and standardizing the acquired image data for image distortion; extracting features for judging the degree of posture abnormality, selecting the coordinates of key human body surface features from multiple angles, and constructing a feature set; extracting joint coordinates using an improved YOLO pose estimation algorithm and a UNet-V2 image segmentation algorithm, and training a posture evaluation model; analyzing the posture images of the subject using the posture evaluation model, and outputting a comprehensive score and evaluation report; periodically updating the model parameters based on accumulated subject data; and remotely storing, managing, and transmitting the data through a network system.

[0005] Based on the above-described method embodiments, the multi-angle posture analysis and evaluation method provided in this embodiment of the invention, wherein the non-contact acquisition of posture data of the front, side, back, and top of the human body using a wide-angle camera includes: installing two wide-angle cameras using a long bracket and a short bracket respectively; the short bracket wide-angle camera is 1.2 meters away from the wall and 1 meter high, used to capture images of the front, side, and back; the long bracket wide-angle camera is 2.2 meters above the ground and extends horizontally for 1.2 meters, used to capture images of the top of the head; the subject completes three cycles of acquisition in a clockwise order of front, side, back, and another side; when acquiring the back image, the subject completes the first set of actions of bowing the head, bending over with hands clasped, and bending horizontally; when acquiring the top of the head image, the subject completes the second set of actions of reaching horizontally and bending over to touch the ground; the first set of actions and the second set of actions are each repeated three times.

[0006] Based on the above method embodiments, the multi-angle body posture analysis and evaluation method provided in this embodiment of the invention includes image distortion correction and standardization of the acquired image data, including: adjusting radial and tangential distortion, eliminating radial and tangential distortion of wide-angle lenses, and performing angle standardization processing on the image to correct shooting angle deviation.

[0007] Based on the above method embodiments, the multi-angle posture analysis and evaluation method provided in this embodiment of the invention extracts the characteristics for judging the degree of posture abnormality, including: head tilt index, shoulder height imbalance index, scapular height imbalance index, iliac crest height imbalance index, spinal coronal plane balance index, arm-to-trunk circumference ratio, and trunk torsion angle. The trunk torsion angle is used to quantify back asymmetry, and the arm-to-trunk circumference ratio is used to assess spinal deviation.

[0008] Based on the above method embodiments, the multi-angle posture analysis and evaluation method provided in this embodiment of the invention, which uses an improved YOLO posture estimation algorithm and a UNet-V2 image segmentation algorithm to extract joint coordinates and train a posture evaluation model, includes: loading an attention mechanism and a target detection head into the YOLOv11 model, replacing the YOLOv11 backbone network with UNet-V2, and optimizing the model parameters through K-fold cross-validation. The YOLOv11 model is used for posture joint capture, and UNet-V2 is used for posture image segmentation.

[0009] Based on the above method embodiments, the multi-angle posture analysis and evaluation method provided in this embodiment of the invention, wherein the posture evaluation model is used to analyze the subject's posture image and output a comprehensive score and evaluation report, includes: using a lightweight Res-Transformer model as a feature extraction network, fusing multi-angle features through a cross-attention mechanism to construct a ReT-CAN multi-angle fusion model; outputting a comprehensive score and a judgment result containing "normal" or "abnormal", wherein the judgment result includes a head flexion judgment result, a trunk rotation judgment result and / or a trunk lateral flexion judgment result.

[0010] Based on the above method embodiments, the multi-angle body posture analysis and evaluation method provided in this embodiment of the invention includes the following: remote storage, management and transmission of data through a network system, including: remotely acquiring the subject's body posture data and comprehensive score, generating an evaluation result that includes a comparison between the ideal value and the actual value of the indicator, feeding back the evaluation result to the subject, and periodically using K-fold cross-validation to validate the updated model parameters for the updated data.

[0011] Secondly, embodiments of the present invention provide a multi-angle posture analysis and evaluation device, comprising: a first main module for non-contact acquisition of posture data of the front, side, back, and top of a human body using a wide-angle camera; a second main module for image distortion correction and standardization of the acquired image data; a third main module for extracting features for judging the degree of posture abnormality, screening the coordinates of key human body surface features from multiple angles, and constructing a feature set; a fourth main module for extracting joint coordinates using an improved YOLO pose estimation algorithm and a UNet-V2 image segmentation algorithm, and training a posture evaluation model; a fifth main module for analyzing the posture images of the subject using the posture evaluation model, and outputting a comprehensive score and evaluation report; a sixth main module for periodically updating model parameters based on accumulated subject data; and a seventh main module for remote storage, management, and transmission of data via a network system.

[0012] Thirdly, embodiments of the present invention provide an electronic device, comprising:

[0013] At least one processor, at least one memory, and a communication interface; wherein,

[0014] The processor, memory, and communication interface communicate with each other;

[0015] The memory stores program instructions that can be executed by the processor. The processor calls the program instructions to execute the multi-angle posture analysis and evaluation method provided by any one of the various implementations of the first aspect.

[0016] Fourthly, embodiments of the present invention provide a non-transitory computer-readable storage medium storing computer instructions that cause a computer to execute a multi-angle body analysis and evaluation method provided by any of the various implementations of the first aspect.

[0017] The multi-angle posture analysis and evaluation method and device provided in this invention achieve multi-angle data acquisition and intelligent evaluation through non-contact posture capture technology. It uses two ordinary wide-angle cameras to capture human posture images from the front, side, back, and top, and performs distortion correction and standardization processing using a self-developed algorithm. A feature set is constructed using multi-dimensional evaluation indicators, joint point coordinates are extracted and the model is trained, and K-fold cross-validation is used for optimization. Multi-angle features are fused to generate a comprehensive score and report, achieving a "normal / abnormal" judgment. Accumulated data is used to periodically update model parameters, and remote data storage, transmission, and management are achieved through a network system, providing an efficient solution for posture analysis. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the 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 based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the multi-angle posture analysis and evaluation method provided in an embodiment of the present invention;

[0020] Figure 2 This is a schematic diagram of the structure of the multi-angle body posture analysis and evaluation device provided in an embodiment of the present invention;

[0021] Figure 3 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention;

[0022] Figure 4 This is a schematic diagram of the interface effect of the body posture data acquisition system provided in an embodiment of the present invention;

[0023] Figure 5 This is a schematic diagram illustrating the dynamic torso twist angle effect provided in an embodiment of the present invention;

[0024] Figure 6 This is a schematic diagram illustrating the effect of the frontal image evaluation index provided in an embodiment of the present invention;

[0025] Figure 7 This is a schematic diagram illustrating the effect of side image metrics provided in an embodiment of the present invention;

[0026] Figure 8This is a schematic diagram illustrating the effect of back image metrics provided in an embodiment of the present invention;

[0027] Figure 9 This is a schematic diagram illustrating the effect of the top-of-the-head image indicators provided in an embodiment of the present invention;

[0028] Figure 10 This is a schematic diagram of the interface effect of the intelligent posture assessment system provided in an embodiment of the present invention;

[0029] Figure 11 This is a schematic diagram of the login page effect of the data storage, management and transmission system provided in an embodiment of the present invention;

[0030] Figure 12 This is a schematic diagram of the main page effect of the data storage, management and transmission system provided in an embodiment of the present invention. Detailed Implementation

[0031] 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. In addition, the technical features of the various embodiments or individual embodiments provided by the present invention can be arbitrarily combined to form feasible technical solutions. Such combinations are not constrained by the order of steps and / or structural composition patterns, but must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention. If there are step numbers in the following embodiments, they are only set for ease of explanation and the order between steps is not limited. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0032] This invention provides a multi-angle posture analysis and evaluation method, see [link to relevant documentation]. Figure 1 The method includes: non-contact acquisition of human posture data from the front, side, back, and top of the body using a wide-angle camera; image distortion correction and standardization of the acquired image data; extraction of features for assessing the degree of posture abnormality, selection of key human body surface feature coordinates from multiple angles, and construction of a feature set; extraction of joint coordinates using an improved YOLO pose estimation algorithm and a UNet-V2 image segmentation algorithm, and training of a posture assessment model; analysis of the subject's posture images using the posture assessment model, and output of a comprehensive score and evaluation report; periodic updating of model parameters based on accumulated subject data; and remote storage, management, and transmission of data through a network system.

[0033] Based on the above method embodiments, as an optional embodiment, the multi-angle posture analysis and evaluation method provided in this embodiment of the invention, wherein the non-contact acquisition of posture data of the front, side, back, and top of the human body using a wide-angle camera includes: installing two wide-angle cameras using a long bracket and a short bracket respectively; the short bracket wide-angle camera is 1.2 meters away from the wall and 1 meter high, used to capture images of the front, side, and back; the long bracket wide-angle camera is 2.2 meters above the ground and extends horizontally for 1.2 meters, used to capture images of the top of the head; the subject completes three cycles of acquisition in a clockwise order of front, side, back, and another side; when acquiring the back image, the subject completes the first set of actions of bowing the head, bending over with hands clasped, and bending horizontally; when acquiring the top of the head image, the subject completes the second set of actions of reaching horizontally and bending over to touch the ground; the first set of actions and the second set of actions are each repeated three times.

[0034] A designated area for posture assessment was established, and the necessary testing equipment was installed. The entire posture assessment environment was located indoors to ensure participant privacy. A diagram illustrating the connection of the posture detection equipment is shown below. Figure 1 As shown, two sets of tripods and two standard wide-angle cameras were used to achieve multi-angle shooting. The short tripod was placed in front of the shooting environment, directly facing the center of the suspended backdrop, with its location marked by yellow warning tape on the ground. The wide-angle camera on this tripod was used to capture front, side, and back images, 1.2 meters from the wall and 1 meter high. The long tripod was located on the left side of the shooting environment, also marked by yellow warning tape on the ground. The wide-angle camera on this tripod was used to capture images above the head, with the device 2.2 meters off the ground and the tripod extending horizontally for 1.2 meters, directly facing the footprints in the shooting area. Both wide-angle cameras were connected to a mobile PC. Self-made image acquisition software was used to capture camera images to obtain high-quality, wide-angle human posture image data. During the shooting process, all lights in the shooting environment were kept on to avoid shadows from curtains and the subject's hair affecting the shooting.

[0035] Body posture data was captured using a standard wide-angle camera, a 4K high-definition wide-angle camera from Hikvision, capable of capturing fine image details. Furthermore, compared to industrial camera lenses, this camera lens offers a wide field of view, covering a broad scene range even with distance limitations, capturing wide-angle images. Secondly, the inventors developed a body posture data acquisition system (such as...). Figure 4 As shown in the figure, the system can save the collected body photos on the computer, and the data is automatically numbered according to different subjects.

[0036] After the equipment is set up, the personnel will ask the subject to stand on the acquisition platform and, according to the required movements, acquire images from four angles in a clockwise sequence. The frontal image is acquired first, followed by the side image, then the back image, and finally another side image before returning to the front, repeating this cycle three times. When acquiring the back image, the subject will be asked to perform three test movements: head down, hands clasped and bending over, and horizontal bending over. A wide-angle camera will be used to acquire the coordinates of key human body surface features (such as upper chest, thoracic, and lumbar regions) from multiple angles. Figure 5 (As shown), the data collection is repeated three times. After completing the above data collection, the camera is switched to the overhead camera to collect images from above. The subject is required to complete two sets of actions: reaching out horizontally and bending over to touch the ground. This process is repeated three times. Each subject undergoes three sets of experiments to reduce errors caused by internal and external factors. The collected data is then used by professionals to select and evaluate body posture data corresponding to the preferred characteristics.

[0037] Based on the above method embodiments, as an optional embodiment, the multi-angle body posture analysis and evaluation method provided in this embodiment of the invention includes image distortion correction and standardization of the acquired image data, including: adjusting radial and tangential distortion, eliminating radial and tangential distortion of wide-angle lenses, and performing angle standardization processing on the image to correct shooting angle deviation.

[0038] The generated dataset is preprocessed using a computer, including data distortion correction and data standardization. First, the distortion parameters radial distortion k_1 and tangential distortion k_2 are adaptively adjusted to minimize radial and tangential distortion caused by the wide-angle camera lens, thereby correcting the image and making it closer to the projection of the real world, thus improving image quality. Then, the images are angle-standardized, shifting any possible shooting angle towards the ideal angle, making the image information more accurate. When shooting human images, there may be slight tilting of the person's position; angle standardization can minimize this tilt.

[0039] Based on the above method embodiments, as an optional embodiment, the multi-angle posture analysis and evaluation method provided in this embodiment of the invention extracts the characteristics for judging the degree of posture abnormality, including: head tilt index, shoulder height imbalance index, scapular height imbalance index, iliac crest height imbalance index, spinal coronal plane balance index, arm-to-trunk circumference ratio, and trunk torsion angle. The trunk torsion angle is used to quantify back asymmetry, and the arm-to-trunk circumference ratio is used to assess spinal deviation.

[0040] Relevant evaluation indicators (as shown in Table 1) were selected by professionals and their correspondences were established with the coordinates of key human body surface features from multiple angles, serving as the screening criteria for training data. The preferred indicators included head tilt index, shoulder height imbalance index, scapular height imbalance index, iliac crest height imbalance index, spinal coronal plane balance index, scapular asymmetry index, iliac crest asymmetry index, arm-to-trunk circumference ratio, pelvic offset, head rotation index, shoulder height imbalance index, thoracic kyphosis angle, forward tilt angle, lateral posture classification, trunk torsion angle, and back torsion ratio, covering four shooting angles for comprehensive evaluation.

[0041] Table 1

[0042]

[0043] Based on the above method embodiments, as an optional embodiment, the multi-angle posture analysis and evaluation method provided in this embodiment of the invention, which uses an improved YOLO pose estimation algorithm and a UNet-V2 image segmentation algorithm to extract joint coordinates and train a posture evaluation model, includes: loading an attention mechanism and a target detection head into the YOLOv11 model, replacing the YOLOv11 backbone network with UNet-V2, and optimizing the model parameters through K-fold cross-validation. The YOLOv11 model is used for posture joint capture, and the UNet-V2 is used for posture image segmentation.

[0044] The coordinates of key points in the image at the corresponding shooting angle are extracted for subsequent training of the deep learning model and to assist in the subsequent digital evaluation. A schematic diagram of the frontal image metrics is shown below. Figure 6 As shown in the schematic diagram of the side view. Figure 7 As shown in the diagram, the back image is as follows: Figure 8 As shown in the diagram, the top of the head is illustrated as follows: Figure 9 As shown, a customized digital intelligent postural anomaly assessment system (such as...) was developed. Figure 10 As shown in the figure, the system can perform feature extraction and subsequent motion evaluation.

[0045] The model training employs an improved YOLO pose estimation algorithm and an improved UNet-V2 image segmentation algorithm. The former is used for pose keypoint capture, and the latter for pose image segmentation. The YOLO-Pose pose estimation model is a sub-model of the larger YOLO model and is integrated into subsequent YOLO versions. YOLOv11 is the latest YOLO series model, boasting high accuracy and low computational cost. This invention improves the pose estimation algorithm of the original YOLOv11 model and modifies the overall model architecture, introducing a new attention mechanism and object detection head. On the other hand, UNet has long demonstrated excellent performance in few-shot segmentation tasks and is widely used. UNet-V2, an improved variant of UNet, is a more efficient variant that maintains good segmentation capabilities while optimizing parameter control. This invention replaces the YOLOv11 backbone with UNet-V2, learning sample features through a lighter model, reducing the risk of overfitting, and enabling its use in pose image segmentation.

[0046] Based on the above method embodiments, as an optional embodiment, the multi-angle posture analysis and evaluation method provided in this embodiment of the invention, wherein the posture evaluation model is used to analyze the subject's posture image and output a comprehensive score and evaluation report, includes: using a lightweight Res-Transformer model as a feature extraction network, fusing multi-angle features through a cross-attention mechanism to construct a ReT-CAN multi-angle fusion model; outputting a comprehensive score and a judgment result containing "normal" or "abnormal", wherein the judgment result includes head flexion judgment result, trunk rotation judgment result and / or trunk lateral flexion judgment result.

[0047] A digital posture assessment algorithm has been developed. In daily use, based on the steps described above, a trained model can acquire the subject's feature set, automatically calculate the degree of deviation from normal posture, and provide a comprehensive evaluation score. This digital posture assessment algorithm is based on the Transformer model. This model architecture abandons the recursive structure of traditional Recurrent Neural Networks (RNNs), achieving parallel computation and higher training efficiency. This invention lightweights the existing Transformer model, constructing a new Res-Transformer model as the feature extraction network. Multiple-angle image data collected by the training model are input into the Res-Transformer, and a cross-attention mechanism is used to fuse the multi-angle features. The fused features are then used for subsequent scoring tasks, ultimately constructing a multi-angle fusion model called ReT-CAN (Res-Transformer-Cross Attention), which can automatically output a comprehensive score and a comprehensive evaluation report.

[0048] Preferably, the digital assessment algorithm employs a binary judgment, meaning the final subject's assessment result is either "normal" or "abnormal." Based on postural spatial position, researchers summarized seven possible final diagnostic opinions based on indicators: head forward flexion, head rotation, head lateral deviation, trunk forward tilt, trunk backward tilt, trunk lateral flexion, and trunk rotation. To improve the accuracy of the assessment model in actual use, the assessment system accumulates the subject's postural data, and k-fold cross-validation is used periodically to train and update the parameters of each assessment model.

[0049] Based on the above method embodiments, as an optional embodiment, the multi-angle body posture analysis and evaluation method provided in this embodiment of the invention, wherein the remote storage, management and transmission of data through a network system includes: remotely acquiring the subject's body posture data and comprehensive score, generating an evaluation result that includes a comparison between the ideal value and the actual value of the indicator, feeding back the evaluation result to the subject, and periodically using K-fold cross-validation to validate the updated model parameters for the updated data.

[0050] Personnel can log in to data storage, management and transmission systems (such as...) Figure 10 (As shown) Remotely acquire subject information, calculate the final score and body posture data as a reference to assess the subject's abnormal body posture, compile a comprehensive assessment report, and remotely send the assessment report to the subject through a data storage management and transmission system.

[0051] The multi-angle posture analysis and evaluation method provided in this invention achieves multi-angle data acquisition and intelligent evaluation through non-contact posture capture technology. It utilizes two ordinary wide-angle cameras to capture human posture images from the front, side, back, and top, and performs distortion correction and standardization processing using a self-developed algorithm. A feature set is constructed using multi-dimensional evaluation indicators, keypoint coordinates are extracted and the model is trained, and K-fold cross-validation is used for optimization. Multi-angle features are fused to generate a comprehensive score and report, achieving a "normal / abnormal" judgment. Accumulated data is used to periodically update model parameters, and remote data storage, transmission, and management are achieved through a network system, providing an efficient solution for posture analysis.

[0052] The implementation of the various embodiments of the present invention is based on programmed processing (i.e., software) using a device with processor functionality. Therefore, in practical engineering, the technical solutions and functions of the various embodiments of the present invention can be encapsulated into various modules. Based on this reality, and building upon the above embodiments, the embodiments of the present invention provide a multi-angle posture analysis and evaluation device, which is used to execute the multi-angle posture analysis and evaluation method described in the above method embodiments. See also... Figure 2The device includes: a first main module for non-contact acquisition of human posture data from the front, side, back, and top of the body using a wide-angle camera; a second main module for image distortion correction and standardization of the acquired image data; a third main module for extracting features for assessing the degree of posture abnormality, selecting the coordinates of key human body surface features from multiple angles, and constructing a feature set; a fourth main module for extracting joint coordinates using an improved YOLO pose estimation algorithm and a UNet-V2 image segmentation algorithm, and training a posture assessment model; a fifth main module for analyzing the subject's posture images using the posture assessment model, and outputting a comprehensive score and evaluation report; a sixth main module for periodically updating model parameters based on accumulated subject data; and a seventh main module for remote storage, management, and transmission of data via a network system.

[0053] The multi-angle body posture analysis and evaluation device provided in this embodiment of the invention adopts... Figure 2 Several modules within the system utilize non-contact posture capture technology to achieve multi-angle data acquisition and intelligent evaluation. Two ordinary wide-angle cameras are used to capture human posture images from the front, side, back, and top, which are then processed and standardized using a self-developed algorithm for distortion correction. A feature set is constructed using multi-dimensional evaluation indicators, keypoint coordinates are extracted and the model is trained, and K-fold cross-validation is employed for optimization. Multi-angle features are fused to generate a comprehensive score and report, enabling "normal / abnormal" assessment. Accumulated data is used to periodically update model parameters, and remote data storage, transmission, and management are achieved through a network system, providing an efficient solution for posture analysis.

[0054] It should be noted that the apparatus in the device embodiments provided by the present invention can be used not only to implement the methods in the above method embodiments, but also to implement the methods in other method embodiments provided by the present invention. The difference lies only in setting corresponding functional modules (i.e., software). Its principle is basically the same as that of the above device embodiments provided by the present invention. As long as those skilled in the art, based on the above device embodiments and referring to the specific technical solutions in other method embodiments, obtain corresponding technical means and technical solutions composed of these technical means by combining technical features, and under the premise of ensuring the practicality of the technical solution, they can improve the apparatus in the above device embodiments to obtain corresponding device-type embodiments (i.e., software) for implementing the methods in other method-type embodiments. For example:

[0055] Based on the above-described device embodiments, as an optional embodiment, the multi-angle posture analysis and evaluation device provided in this embodiment of the invention further includes: a first sub-module, used to realize the non-contact acquisition of posture data of the front, side, back, and top of the human body using a wide-angle camera, including: using a long bracket and a short bracket to install two wide-angle cameras respectively, the short bracket wide-angle camera being 1.2 meters away from the wall and 1 meter high, used to capture front, side, and back images; the long bracket wide-angle camera being 2.2 meters above the ground and extending horizontally for 1.2 meters, used to capture images of the top of the head; the subject completes three cycles of acquisition in a clockwise order of front, side, back, and another side, the back acquisition completing the first set of actions of bowing the head, bending over with hands clasped, and bending horizontally, the top acquisition completing the second set of actions of reaching horizontally and bending over to touch the ground, the first set of actions and the second set of actions are each repeated three times.

[0056] Based on the above-described device embodiments, as an optional embodiment, the multi-angle body posture analysis and evaluation device provided in this embodiment of the invention further includes: a second sub-module, used to perform image distortion correction and standardization on the acquired image data, including: adjusting radial distortion and tangential distortion, eliminating radial distortion and tangential distortion of wide-angle lenses, and performing angle standardization processing on the image to correct shooting angle deviation.

[0057] Based on the above-described device embodiments, as an optional embodiment, the multi-angle posture analysis and evaluation device provided in this embodiment of the invention further includes: a third sub-module, used to extract the characteristics for judging the degree of posture abnormality, including: head tilt index, shoulder height imbalance index, scapular height imbalance index, iliac crest height imbalance index, spinal coronal plane balance index, arm-to-trunk circumference ratio, and trunk torsion angle, wherein the trunk torsion angle is used to quantify back asymmetry, and the arm-to-trunk circumference ratio is used to assess spinal deviation.

[0058] Based on the above-described device embodiments, as an optional embodiment, the multi-angle posture analysis and evaluation device provided in this embodiment of the invention further includes: a fourth sub-module, used to implement the extraction of joint coordinates using the improved YOLO posture estimation algorithm and the UNet-V2 image segmentation algorithm, and to train the posture evaluation model, including: loading an attention mechanism and a target detection head into the YOLOv11 model, replacing the YOLOv11 backbone network with UNet-V2, and optimizing the model parameters through K-fold cross-validation, wherein the YOLOv11 model is used for posture joint capture, and the UNet-V2 is used for posture image segmentation.

[0059] Based on the above-described device embodiments, as an optional embodiment, the multi-angle posture analysis and evaluation device provided in this embodiment of the invention further includes: a fifth sub-module, used to implement the analysis of the subject's posture image using the posture evaluation model, and output a comprehensive score and evaluation report, including: using a lightweight Res-Transformer model as a feature extraction network, fusing multi-angle features through a cross-attention mechanism to construct a ReT-CAN multi-angle fusion model; outputting a comprehensive score and a judgment result containing "normal" or "abnormal", wherein the judgment result includes a head flexion judgment result, a trunk rotation judgment result, and / or a trunk lateral flexion judgment result.

[0060] Based on the above-described device embodiments, as an optional embodiment, the multi-angle posture analysis and evaluation device provided in this embodiment of the invention further includes: a sixth sub-module, used to realize the remote storage, management and transmission of data through the network system, including: remotely acquiring the subject's posture data and comprehensive score, generating an evaluation result that includes a comparison between the ideal value and the actual value of the indicator, and feeding the evaluation result back to the subject, and periodically using K-fold cross-validation to validate the updated model parameters for the updated data.

[0061] The method in this embodiment of the invention is implemented using an electronic device; therefore, it is necessary to introduce the relevant electronic device. For this purpose, this embodiment of the invention provides an electronic device, such as... Figure 3 As shown, the electronic device includes at least one processor, a communications interface, at least one memory, and a communications bus, wherein the at least one processor, the communications interface, and the at least one memory communicate with each other via the communications bus. The at least one processor can invoke logical instructions stored in the at least one memory to execute all or part of the steps of the methods provided in the foregoing method embodiments.

[0062] Furthermore, when the logical instructions in at least one of the aforementioned memories can be 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, in essence, or the part that contributes to the prior art, or a part 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 to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various method embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0063] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0064] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0065] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. Based on this understanding, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, or sometimes in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0066] It should be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. Any expressions such as "predetermined threshold," "preset threshold," etc., without specifying a particular value, can be determined by those skilled in the art through simple experimentation or appropriate adjustments.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-angle posture analysis and evaluation method, characterized in that, include: Human body posture data is collected non-contactly from the front, side, back, and top using a wide-angle camera; The acquired image data undergoes image distortion correction and standardization. From the distorted and standardized body posture data, features for assessing the degree of postural abnormality are selected. Key human body surface feature coordinates from multiple angles are screened, relevant evaluation indicators are chosen, and a correspondence is established between these indicators and the key human body surface feature coordinates from multiple angles. This correspondence serves as the selection criterion for training data, constructing a feature set. A modified YOLO pose estimation algorithm and a UNet-V2 image segmentation algorithm are used to extract joint coordinates, training the body posture assessment model. The body posture assessment model is used to analyze the subject's body posture images, outputting a comprehensive score and evaluation report. Model parameters are updated periodically based on accumulated subject data. Remote storage, management, and transmission of data are achieved through a network system. The characteristics for assessing the degree of postural abnormality include: head tilt index, shoulder height imbalance index, scapular height imbalance index, iliac crest height imbalance index, spinal coronal plane balance index, arm-to-trunk circumference ratio, and trunk torsion angle. The trunk torsion angle is used to quantify back asymmetry, and the arm-to-trunk circumference ratio is used to assess spinal deviation.

2. The multi-angle posture analysis and evaluation method according to claim 1, characterized in that, The method of non-contactly collecting body posture data of the front, side, back, and top of the human body using a wide-angle camera includes: mounting two wide-angle cameras on a long stand and a short stand respectively. The wide-angle camera on the short stand is 1.2 meters away from the wall and 1 meter high, used to capture images of the front, side, and back. The wide-angle camera on the long stand is 2.2 meters above the ground and extends horizontally for 1.2 meters, used to capture images of the top of the head. The subject completes three cycles of collecting images of the front, side, back, and another side in a clockwise order. When collecting images of the back, the subject completes the first set of actions: bending down, clasping hands, and bending horizontally. When collecting images of the top of the head, the subject completes the second set of actions: reaching horizontally and bending over to touch the ground. The first and second sets of actions are each repeated three times.

3. The multi-angle posture analysis and evaluation method according to claim 2, characterized in that, The image distortion correction and standardization of the acquired image data includes: adjusting radial and tangential distortion, eliminating radial and tangential distortion of wide-angle lenses, and performing angle standardization processing on the image to correct shooting angle deviation.

4. The multi-angle posture analysis and evaluation method according to claim 3, characterized in that, The method of extracting joint coordinates using an improved YOLO pose estimation algorithm and a UNet-V2 image segmentation algorithm and training a body posture evaluation model includes: loading an attention mechanism and a target detection head into the YOLOv11 model, replacing the YOLOv11 backbone network with UNet-V2, and optimizing model parameters through K-fold cross-validation. The YOLOv11 model is used for pose joint capture, and UNet-V2 is used for pose image segmentation.

5. The multi-angle posture analysis and evaluation method according to claim 4, characterized in that, The step of analyzing the subject's posture images using the posture assessment model and outputting a comprehensive score and evaluation report includes: using a lightweight Res-Transformer model as the feature extraction network, fusing multi-angle features through a cross-attention mechanism, and constructing a ReT-CAN multi-angle fusion model; outputting a comprehensive score and a judgment result containing "normal" or "abnormal", wherein the judgment result includes head flexion judgment result, trunk rotation judgment result and / or trunk lateral flexion judgment result.

6. The multi-angle posture analysis and evaluation method according to claim 5, characterized in that, The remote storage, management and transmission of data through the network system includes: remotely acquiring the subject's body posture data and comprehensive score, generating an evaluation result that includes a comparison between the ideal value and the actual value of the indicator, feeding the evaluation result back to the subject, and periodically using K-fold cross-validation to validate the updated model parameters for the updated data.

7. A multi-angle posture analysis and evaluation device for implementing the multi-angle posture analysis and evaluation method as described in any one of claims 1 to 6, characterized in that, include: The first main module is used to realize the non-contact acquisition of human body posture data from the front, side, back and top of the human body through a wide-angle camera; The second main module is used to perform image distortion correction and standardization on the acquired image data; The third main module is used to select features for judging the degree of postural abnormality from the postural data after distortion correction and standardization, filter the required multi-angle key human body surface feature coordinates, select relevant evaluation indicators, and establish a correspondence between them and the multi-angle key human body surface feature coordinates as the selection criteria for training data, thus constructing a feature set. The fourth main module is used to extract joint coordinates using an improved YOLO pose estimation algorithm and a UNet-V2 image segmentation algorithm to train the postural assessment model. The fifth main module is used to analyze the subject's postural images using the postural assessment model and output a comprehensive score and evaluation report. The sixth main module is used to periodically update the model parameters based on the accumulated subject data. The seventh main module is used to realize the remote storage, management and transmission of data through the network system; The characteristics for assessing the degree of postural abnormality include: head tilt index, shoulder height imbalance index, scapular height imbalance index, iliac crest height imbalance index, spinal coronal plane balance index, arm-to-trunk circumference ratio, and trunk torsion angle. The trunk torsion angle is used to quantify back asymmetry, and the arm-to-trunk circumference ratio is used to assess spinal deviation.

8. An electronic device, characterized in that, include: At least one processor, at least one memory, and a communication interface; wherein, The processor, memory, and communication interface communicate with each other; The memory stores program instructions that can be executed by the processor, which invokes the program instructions to perform the method described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores computer instructions that cause the computer to perform the method described in any one of claims 1 to 6.

Images