Human meridian and acupoint positioning method and device, electronic equipment and medium

Abstract

Embodiments of the present application provide a human meridian and acupoint positioning method and device, electronic equipment and medium, comprising: collecting single-view global and / or local target human body data of a target object; inputting the target human body data into a pre-trained human acupoint positioning network model for human acupoint labeling; positioning the human meridian and acupoint of the target object based on the human acupoint labeling result and the meridian and vessel direction of the human body; the human acupoint positioning network model generates a standard three-dimensional human body model by constructing a first parameterized human body model of a standard posture and performing grid refinement, labels acupoint points on the surface vertices of the model, and constructs a standard three-dimensional human acupoint model; based on the standard three-dimensional human acupoint model and a plurality of single-view global and / or local parameterized human body data, the model is trained to obtain a human acupoint positioning network model. Therefore, the positions of acupoints and meridians can be found more accurately, quickly and intuitively, so as to learn the meridian and acupoint system more accurately.

Description

Technical Field

[0001] This invention relates to the field of human acupoint location and three-dimensional data processing technology, and in particular to a method, device, electronic device and medium for locating human meridian acupoints. Background Technology

[0002] Acupoints, also known as acupuncture points, are locations on the body surface where the Qi (vital energy) of the internal organs and meridians is transported. Current acupoint location methods involve using a handheld scanning device to scan a standard human body to create a three-dimensional human body model. Acupoints are then established on this model to construct a three-dimensional meridian and acupoint model. Scanning data of surface landmarks from a patient's three-dimensional body surface structure model is then acquired and matched to the patient's model, allowing for real-time display of the patient's three-dimensional meridian and acupoint system structure.

[0003] However, existing methods for locating acupoints, whether using 3D data of a standard human body or a patient's body, require the use of a handheld high-precision 3D scanner. This equipment is not only expensive but also inconvenient to use. Furthermore, different people have different heights, weights, and subcutaneous tissue distributions. In addition, traditional Chinese medicine acupuncture emphasizes the sensation of "deqi" (the feeling of obtaining qi), making it difficult to represent acupoints with a fixed needling depth. When matching a 3D meridian and acupoint model created on a standard human body to a 3D body surface structure model of a patient, multiple surface markers need to be pre-marked on the 3D human body model to complete the model matching. Since the human body is a non-rigid hinge structure with various structural attributes such as height, weight, and body shape, it is difficult to achieve an accurate match between two models by simply stretching or shrinking them using a few dozen surface markers. Summary of the Invention

[0004] This invention provides a method, device, electronic device, and medium for locating acupoints and meridians in the human body. It addresses the shortcomings of existing technologies that require the use of a handheld high-precision 3D scanner to obtain three-dimensional human body data and rely solely on stretching or reducing the three-dimensional meridian and acupoint model and the patient's three-dimensional body surface structure model using only a few dozen surface markers, making it difficult to achieve a precise match between the two models. This invention enables accurate, rapid, and intuitive location of meridians and acupoints.

[0005] This invention provides a method for locating acupoints along human meridians, comprising:

[0006] Collect single-view global and / or local target human body data of the target object;

[0007] The target human body data is input into a pre-trained human acupoint location network model, and the human acupoint location network model is used to annotate the target human body data to obtain the human acupoint annotation results.

[0008] Based on the aforementioned human acupoint annotation results and the direction of human meridians, the human meridian acupoints of the target object are located;

[0009] The human acupoint location network model is trained based on the following steps:

[0010] A first parametric human body model with a standard pose is constructed, and the mesh of the first parametric human body model is refined to generate a standard three-dimensional human body model. The number of surface points in the standard three-dimensional human body model is greater than the number of surface points in the parametric human body model.

[0011] Acupoints are marked on the vertices of the body surface of the standard three-dimensional human body model to construct a standard three-dimensional human acupoint model;

[0012] Acquire global and / or local parametric human body data from multiple single-viewpoints, and generate a three-dimensional human acupoint model corresponding to each single-viewpoint global and / or local parametric human body data based on the standard three-dimensional human acupoint model.

[0013] The neural network model is trained based on the global and / or local parameterized human body data under multiple single-view perspectives and the three-dimensional human acupoint model corresponding to the global and / or local parameterized human body data under each single-view perspective, to obtain the human acupoint positioning network model.

[0014] In one possible implementation, the method further includes:

[0015] Based on the human acupoint marking results and the direction of human meridians, two adjacent acupoints are connected to obtain multiple line segments between acupoints;

[0016] Multiple positioning points are located on each acupoint line segment, and the nearest neighbor search method is used to find the nearest neighbor of each positioning point on the target human body data;

[0017] By assigning a corresponding color to the nearest neighbor of each location point based on different meridians, the location of the human meridian acupoints corresponding to the target object can be obtained.

[0018] In one possible implementation, the method further includes:

[0019] Construct second parametric human body models with different postures and body types;

[0020] The second parametric human body model is refined into a mesh of the same standard as the standard three-dimensional human acupoint model to generate a three-dimensional human body model corresponding to the second parametric human body model.

[0021] By annotating the acupoints in the standard three-dimensional human acupoint model, the three-dimensional human acupoint model is obtained by annotating the acupoints in the three-dimensional human acupoint model.

[0022] The position of the virtual camera is randomly set, and global and / or local human body data of the three-dimensional human body model under multiple single views are obtained through the virtual camera, as well as the three-dimensional human body acupoint model corresponding to the global and / or local human body data under each single view.

[0023] In one possible implementation, the method further includes:

[0024] The global and / or local parameterized human body data under multiple single-view perspectives and the three-dimensional human acupoint model corresponding to each global and / or local parameterized human body data under each single-view perspective are input into the neural network model. The global and / or local parameterized human body data under multiple single-view perspectives are used as the model input data, and the three-dimensional human acupoint model corresponding to each global and / or local parameterized human body data under each single-view perspective is used as the ground truth to train the neural network model.

[0025] During the training of the neural network model, the network parameters are adjusted using an error backpropagation algorithm.

[0026] When the neural network model is trained to the convergence state, it is determined that the training of the neural network model is complete, and the human acupoint positioning network model is obtained.

[0027] In one possible implementation, the method further includes:

[0028] Based on the prescribed standard human acupoint location method and acupoint names, acupoints are marked on the vertices of the standard three-dimensional human body model to construct a standard three-dimensional human acupoint model.

[0029] In one possible implementation, the method further includes:

[0030] The location results of the human meridians and acupoints corresponding to the target object are displayed on an external monitor.

[0031] In one possible implementation, the method further includes:

[0032] The target object's global and / or local human body data is acquired from a single perspective using the camera built into the mixed reality device.

[0033] The present invention also provides a human meridian acupoint positioning device, comprising the following modules:

[0034] The acquisition module is used to acquire single-view global and / or local human body data of the target object;

[0035] The annotation module is used to input the target human body data into a pre-trained human acupoint location network model, and to annotate the target human body data with human acupoints through the human acupoint location network model to obtain human acupoint annotation results.

[0036] The positioning module is used to locate the acupoints of the target object based on the human acupoint annotation results and the direction of human meridians;

[0037] The model training module is used to construct a first parametric human body model in a standard pose, refine the mesh of the first parametric human body model to generate a standard 3D human body model, wherein the number of surface points in the standard 3D human body model is greater than the number of surface points in the parametric human body model; mark acupoints on the vertices of the standard 3D human body model to construct a standard 3D human acupoint model; acquire global and / or local parametric human body data from multiple single-viewpoints, and generate 3D human acupoint models corresponding to the global and / or local parametric human body data from each single-viewpoint based on the standard 3D human acupoint model; train the neural network model based on the global and / or local parametric human body data from multiple single-viewpoints and the 3D human acupoint models corresponding to the global and / or local parametric human body data from each single-viewpoint, to obtain a human acupoint localization network model.

[0038] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the human meridian acupoint location method as described above.

[0039] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the human meridian acupoint location method as described above.

[0040] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the human meridian acupoint location method as described above.

[0041] The present invention provides a method, device, electronic device, and medium for locating acupoints along human meridians. This involves: acquiring single-view global and / or local human body data of a target object; inputting the target human body data into a pre-trained human acupoint location network model; using the human acupoint location network model to annotate the target human body data with acupoints to obtain acupoint annotation results; and locating the acupoints along human meridians of the target object based on the acupoint annotation results and the direction of human meridians. The human acupoint location network model is trained based on the following steps: constructing a first parametric human body model in a standard posture, and refining the mesh of the first parametric human body model to generate a standard three-dimensional model. A standard 3D human body model is constructed, wherein the number of surface points in the standard 3D human body model is greater than the number of surface points in the parameterized human body model. Acupoints are marked on the vertices of the standard 3D human body model to construct a standard 3D human acupoint model. Global and / or local parameterized human body data from multiple single-viewpoints are acquired, and 3D human acupoint models corresponding to the global and / or local parameterized human body data from each single-viewpoint are generated based on the standard 3D human acupoint model. A neural network model is trained based on the global and / or local parameterized human body data from multiple single-viewpoints and the 3D human acupoint models corresponding to the global and / or local parameterized human body data from each single-viewpoint to obtain a human acupoint localization network model. Compared to existing technologies that require a handheld high-precision 3D scanner to obtain three-dimensional human body data and rely solely on stretching or shrinking a three-dimensional meridian and acupoint model and a patient's three-dimensional body surface structure model using only a few dozen surface markers, making it difficult to achieve a precise match between the two models, this solution can more accurately, quickly, and intuitively locate the meridians and acupoints. This allows for a more accurate study of the meridian and acupoint system, and consequently, more precise and effective implementation of traditional Chinese medicine treatments such as acupuncture and massage. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in this 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 this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0043] Figure 1 This is a flowchart illustrating the method for locating acupoints along human meridians provided by the present invention.

[0044] Figure 2 This is a flowchart illustrating the training method for the human acupoint location network model provided by the present invention.

[0045] Figure 3 This is one of the schematic diagrams of the location results of human meridians and acupoints provided by the present invention.

[0046] Figure 4 This is the second schematic diagram of the human meridian and acupoint location results provided by the present invention.

[0047] Figure 5 This is a schematic diagram of a standard three-dimensional human acupoint model provided by the present invention.

[0048] Figure 6 This is a schematic diagram of the second parametric human body model with different postures and body shapes provided by the present invention.

[0049] Figure 7 This is a schematic diagram of the structure of the human meridian and acupoint positioning device provided by the present invention.

[0050] Figure 8 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0052] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of the present invention.

[0053] Figure 1 This is a flowchart illustrating the method for locating acupoints along human meridians provided by the present invention, as shown below. Figure 1 As shown, the method includes the following:

[0054] S11. Collect single-view global and / or local target human body data of the target object.

[0055] This invention is primarily applicable to scenarios involving the location of acupoints and meridians in the human body, such as traditional Chinese medicine acupuncture, treatment, scientific research, and popular science. Users acquire single-view global and / or local target human body data from their own perspective using the RGBD camera built into a mixed reality device. The user can be a doctor, a patient's family member, or a practitioner in a traditional Chinese medicine treatment facility, and the target object is the patient. The mixed reality device can be a head-mounted mixed reality device or mixed reality glasses. The target human body data can be a real-time single-view RGB and depth image of the target object obtained from the user's perspective, and a target human body model can be obtained from this image analysis. Acquiring human body data using the RGBD camera built into the mixed reality device eliminates the need for additional 3D data acquisition equipment, significantly improving ease of use, real-time performance, and reducing costs.

[0056] S12. Input the target human body data into a pre-trained human acupoint positioning network model, and use the human acupoint positioning network model to annotate the target human body data with human acupoints to obtain the human acupoint annotation results.

[0057] In this embodiment of the invention, a human acupoint localization network model is pre-trained. This model has the ability to transform obtained global and / or local single-view human data onto a standard human model. By inputting the target human data into the pre-trained human acupoint localization network model, the acupoints marked on the standard human model can be mapped in real time to the obtained single-view global and / or local human data of the target object. Then, the acupoint localization of the target object is constructed on the obtained global and / or local single-view human data of the target object, resulting in the human acupoint annotation results. The specific model training process is described in... Figure 2 The corresponding embodiments are described in detail, and will not be elaborated here.

[0058] S13. Based on the human acupoint annotation results and the direction of human meridians, locate the human meridians and acupoints of the target object.

[0059] Furthermore, based on the acupoint annotation results of the above model and the existing standard fourteen meridians of the human body, adjacent acupoints are connected to obtain multiple line segments between acupoints; multiple positioning points are taken on each acupoint line segment, and the nearest neighbor search method is used to find the nearest neighbor of each positioning point on the single-view global and / or local human body data of the target object; different colors are preset for each meridian, and the corresponding colors are set for the nearest neighbors of each found positioning point based on different meridians, thus obtaining the human meridian acupoint positioning of the target object, such as... Figure 3 and Figure 4 As shown.

[0060] Specifically, following the pathways of the fourteen meridians, adjacent acupoints are connected in three-dimensional space to obtain line segments. Then, points are taken at sufficiently small intervals along these line segments. The nearest neighbor search method is used to find the nearest neighbor points on the global and / or local single-view human body data of the target object. These points are then assigned different colors according to different meridians. This allows for the real-time generation of a three-dimensional meridian and acupoint model of the target object's human body surface from the user's perspective, displayed on the virtual reality device. As the user's viewpoint moves, the user can locate the three-dimensional acupoint model of the target object's human body surface in real time from their current perspective.

[0061] Furthermore, an external display can be connected to locate and display the corresponding meridians and acupoints of the target object on the external display, so that others other than the user can see the three-dimensional model of the patient's meridians and acupoints on the surface of the body in real time.

[0062] The method for locating acupoints along human meridians provided by this invention involves: collecting single-view global and / or local human body data of a target object; inputting the target human body data into a pre-trained human acupoint location network model; using the human acupoint location network model to annotate the target human body data with acupoints to obtain acupoint annotation results; and locating the acupoints along human meridians of the target object based on the acupoint annotation results and the direction of human meridians. The human acupoint location network model is trained based on the following steps: constructing a first parametric human body model in a standard pose, and refining the mesh of the first parametric human body model to generate a standard three-dimensional human body model. The standard 3D human body model has a greater number of surface points than the parameterized human body model. Acupoints are marked on the vertices of the standard 3D human body model to construct a standard 3D human acupoint model. Multiple global and / or local parameterized human body data from single-viewpoints are acquired, and a 3D human acupoint model corresponding to each of the global and / or local parameterized human body data from a single-viewpoint is generated based on the standard 3D human acupoint model. A neural network model is trained based on the multiple global and / or local parameterized human body data from single-viewpoints and the 3D human acupoint models corresponding to each of the global and / or local parameterized human body data from a single-viewpoint, resulting in a human acupoint localization network model. Compared to existing technologies that require a handheld high-precision 3D scanner to obtain three-dimensional human body data and rely solely on stretching or shrinking a three-dimensional meridian and acupoint model and a patient's three-dimensional body surface structure model using only a few dozen surface markers, making it difficult to achieve a precise match between the two models, this method can more accurately, quickly, and intuitively locate acupoints and meridians. This allows for a more accurate acquisition of the meridian and acupoint system, and thus more precise and effective implementation of traditional Chinese medicine treatments such as acupuncture and massage.

[0063] Figure 2This is a flowchart illustrating the training method for the human acupoint location network model provided by the present invention. The method includes the following:

[0064] S21. Construct a first parametric human body model in a standard pose, and refine the mesh of the first parametric human body model to generate a standard three-dimensional human body model.

[0065] In this embodiment of the invention, a first parametric human model with a standard pose is constructed in a parametric human model generation system. The parametric human model generation system can generate a human model simply by setting human body shape and posture parameters, without the need for additional scanning or acquisition of standard human models defined by other rules.

[0066] Furthermore, to improve the accuracy of acupoint annotation in the standard human body model, the first parametric human body model undergoes multiple mesh refinements to increase the number of surface points, generating a standard 3D human body model. Mesh refinement, in computer graphics, refers to the process of increasing the number of mesh points on the model's surface to enhance model detail. This process can be achieved using various algorithms, such as Loop subdivision, Catmull-Clark subdivision, or Doo-Sabin subdivision. Multiple mesh refinements result in a smoother surface and richer details in the human body model. Increasing the number of surface points in the human body model improves its resolution, allowing for more accurate simulation of subtle human structures such as muscles and skin folds. This is particularly important for acupoint annotation, as the location of acupoints typically requires very precise positioning.

[0067] S22. Mark acupoints on the vertices of the standard three-dimensional human body model to construct a standard three-dimensional human acupoint model.

[0068] On the vertices of the refined standard 3D human body model, each acupoint can be precisely labeled according to the positioning rules of traditional Chinese medicine acupoints. By labeling acupoints on the vertices of the human body model, a complete standard 3D human acupoint model can be constructed. It should be noted that the standard 3D human acupoint model includes multiple acupoint positioning point data, i.e., acupoint coordinate data, such as... Figure 4 As shown in the image, this model not only includes the geometric information of the human body but also the precise location of acupoints, providing an intuitive tool for diagnosis, treatment, and teaching in Traditional Chinese Medicine.

[0069] S23. Obtain global and / or local parametric human body data from multiple single-viewpoints, and generate a three-dimensional human acupoint model corresponding to each single-viewpoint global and / or local parametric human body data based on the standard three-dimensional human acupoint model.

[0070] Multiple second-parameterized human models with different poses and body shapes are constructed in a parametric human model generation system. Furthermore, the meshes of these second-parameterized human models are refined according to the same mesh refinement criteria as the standard 3D human acupoint model, resulting in a corresponding 3D human model for each second-parameterized human model, such as... Figure 6 As shown.

[0071] Furthermore, the acupoints of the three-dimensional human body model obtained above are annotated using the acupoint annotation data in the standard three-dimensional human body acupoint model to obtain the three-dimensional human body acupoint model corresponding to the three-dimensional human body model.

[0072] It should be noted that a standard 3D human acupoint model includes acupoint location data, i.e., acupoint coordinate data. For example, if a standard 3D human body model is labeled with 700 acupoints, then the standard 3D human acupoint model includes the label number of these 700 acupoints and the coordinate data of each acupoint.

[0073] Furthermore, the position of the virtual camera is randomly set, and global and / or local human body data from multiple single-viewpoints of the 3D human body model is acquired through the virtual camera. Ultimately, this yields multiple single-viewpoints of global and / or local human body data, as well as a corresponding 3D human acupoint model for each single-viewpoint.

[0074] For example, a parametric human model generation system can construct 100,000 second-parametric human models with different poses and body shapes, refine the mesh, and obtain a corresponding 3D human model for each second-parametric human model. Then, acupoint annotation can be performed to obtain a 3D human acupoint model corresponding to the 3D human model. Next, by randomly setting the position of a virtual camera in each second-parametric human model scene, 100,000 global and / or local parametric human data points from a single viewpoint can be obtained. Alternatively, by randomly setting the positions of five virtual cameras in each second-parametric human model scene, 500,000 global and / or local parametric human data points from a single viewpoint can be obtained. Furthermore, for example, if a virtual camera captures data of the back of a 3D human model, the acupoint data on the back (including acupoint numbers and coordinates of each acupoint) can be obtained based on the corresponding 3D human acupoint model.

[0075] S24. The neural network model is trained based on the global and / or local parameterized human body data under multiple single views and the three-dimensional human acupoint model corresponding to the global and / or local parameterized human body data under each single view to obtain the human acupoint positioning network model.

[0076] The obtained global and / or local parametric human body data from multiple single-view perspectives, along with the corresponding 3D acupoint models, are input into a neural network model. The model is trained using the global and / or local parametric human body data from multiple single-view perspectives as input data and the corresponding 3D acupoint models as ground truth. The final model can identify and locate acupoints on the global and / or local parametric human body data from single-view perspectives. During the neural network model training process, network parameters are adjusted using an error backpropagation algorithm. When the neural network model reaches convergence, the training is considered complete, resulting in the human acupoint localization network model.

[0077] The criteria for determining model training convergence are as follows:

[0078] Changes in the loss function: Observing the changes in the loss values ​​on the training and validation sets is the most intuitive way to determine whether the model has converged. If the training loss continues to decrease and the validation loss also decreases, it indicates that the model is still learning; if the training loss decreases while the validation loss stabilizes or begins to rise, it may indicate that the model is beginning to overfit; if both the training and validation losses tend to stabilize and their values ​​are not significantly different, it may indicate that the model has converged.

[0079] Training curves: By plotting the training and validation losses as a function of time (or number of iterations), you can more intuitively judge the convergence of the model. If the curve tends to be flat, it usually means that the model has converged.

[0080] Overfitting vs. Underfitting: When overfitting occurs, the model performs well on the training set but poorly on the validation set; when underfitting occurs, the model has a high loss on the training set, and the loss value remains high at the end of training, with large fluctuations in both training and validation losses.

[0081] Performance on the validation set: If the model's performance on the validation set (such as accuracy, F1 score, etc.) no longer improves, or improves very slowly, it can also be used as a signal that the model has converged.

[0082] Learning rate adjustment: If the learning rate has been adjusted to a very small value, but the model's performance still does not improve significantly, it may mean that the model has approached or reached convergence.

[0083] Vanishing or exploding gradients: If the gradient of the model is very small (close to 0), it may cause the gradient to vanish, the model to update very slowly, and it may appear to "converge" but in fact it has not really learned anything; if the gradient is very large, it may cause the gradient to explode, the model parameters to be updated too much, and the loss value may grow explosively.

[0084] In summary, model convergence can be determined by observing the trend of the loss value, the performance on the validation set, and the magnitude of the gradient. Generally, when both the training and validation losses stabilize and the model's performance on the validation set no longer shows significant improvement, the model can be considered to have converged.

[0085] The model has the ability to deform the obtained global and / or local single-view human body data onto a standard human body model.

[0086] This invention utilizes the structured characteristics of parametric human body models to achieve real-time location and display of acupoints and meridians for target objects of any posture and body type from any viewpoint using only a single annotation of acupoints throughout the body. This solves the problems of inaccurate location using rigid stretching and scaling transformations and the dependence of traditional neural networks on massive amounts of labeled data. Furthermore, it utilizes mixed reality devices to collect human body data and display acupoints and meridians on the human body in real time, avoiding the inability of traditional acupoint location methods to adapt to changes in motion. This method for locating and displaying human meridians and acupoints can help individuals and practitioners who need to learn about meridians and acupoints to reduce the learning curve, enabling them to more accurately, quickly, and intuitively locate acupoints and meridians. This facilitates more accurate learning of the meridian and acupoint system and more precise and effective implementation of traditional Chinese medicine treatments such as acupuncture and massage.

[0087] The following describes the human meridian acupoint positioning device provided by the present invention. The human meridian acupoint positioning device described below can be referred to in correspondence with the human meridian acupoint positioning method described above.

[0088] Figure 7 This is a schematic diagram of the structure of the human meridian and acupoint positioning device provided by the present invention, specifically including:

[0089] The acquisition module 701 is used to acquire single-view global and / or local target human body data of the target object. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0090] The annotation module 702 is used to input the target human body data into a pre-trained human acupoint localization network model, and to annotate the target human body data with human acupoints through the human acupoint localization network model to obtain the human acupoint annotation results. For detailed explanations, please refer to the relevant descriptions in the above method embodiments, which will not be repeated here.

[0091] The positioning module 703 is used to locate the acupoints of the target object based on the acupoint marking results and the meridian pathways. For detailed explanation, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0092] The model training module 704 is used to construct a first parametric human body model in a standard pose, and to refine the mesh of the first parametric human body model to generate a standard three-dimensional human body model. The number of surface points in the standard three-dimensional human body model is greater than the number of surface points in the parametric human body model. Acupoints are marked on the vertices of the standard three-dimensional human body model to construct a standard three-dimensional human acupoint model. Multiple global and / or local parametric human body data from single-viewpoints are acquired, and a three-dimensional human acupoint model corresponding to each of the global and / or local parametric human body data from single-viewpoints is generated based on the standard three-dimensional human acupoint model. A neural network model is trained based on the multiple global and / or local parametric human body data from single-viewpoints and the corresponding three-dimensional human acupoint models to obtain a human acupoint localization network model. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0093] Figure 8 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 8 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other through the communication bus 840. The processor 810 can call logical instructions in the memory 830 to execute a method for locating acupoints along human meridians. This method includes: acquiring single-view global and / or local target human body data of a target object; inputting the target human body data into a pre-trained human acupoint location network model, and using the human acupoint location network model to annotate the target human body data with acupoints to obtain acupoint annotation results; and locating the target object's meridians and acupoints based on the acupoint annotation results and the direction of human meridians. The human acupoint location network model is trained based on the following steps: constructing a first parametric human body model in a standard pose, and refining the mesh of the first parametric human body model. A standard 3D human body model is generated, wherein the number of surface points in the standard 3D human body model is greater than the number of surface points in the parameterized human body model. Acupoints are marked on the vertices of the standard 3D human body model to construct a standard 3D human acupoint model. Global and / or local parameterized human body data from multiple single-viewpoints are acquired, and 3D human acupoint models corresponding to the global and / or local parameterized human body data from each single-viewpoint are generated based on the standard 3D human acupoint model. A neural network model is trained based on the global and / or local parameterized human body data from multiple single-viewpoints and the 3D human acupoint models corresponding to the global and / or local parameterized human body data from each single-viewpoint to obtain a human acupoint localization network model.

[0094] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, 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 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.

[0095] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the human meridian acupoint location method provided by the above methods. The method includes: collecting single-view global and / or local target human body data of a target object; inputting the target human body data into a pre-trained human acupoint location network model, and using the human acupoint location network model to annotate the target human body data with human acupoints to obtain human acupoint annotation results; and locating human meridian acupoints of the target object based on the human acupoint annotation results and the direction of human meridians; wherein the human acupoint location network model is trained based on the following steps: constructing... A first parametric human body model in a standard pose is generated by refining the mesh of the first parametric human body model to produce a standard 3D human body model. The number of surface points in the standard 3D human body model is greater than the number of surface points in the parametric human body model. Acupoints are marked on the vertices of the body surface of the standard 3D human body model to construct a standard 3D human acupoint model. Global and / or local parametric human body data from multiple single-viewpoints are acquired, and 3D human acupoint models corresponding to the global and / or local parametric human body data from each single-viewpoint are generated based on the standard 3D human acupoint model. A neural network model is trained based on the global and / or local parametric human body data from multiple single-viewpoints and the 3D human acupoint models corresponding to the global and / or local parametric human body data from each single-viewpoint to obtain a human acupoint localization network model.

[0096] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the human meridian acupoint localization method provided by the above methods. This method includes: acquiring single-view global and / or local target human body data of a target object; inputting the target human body data into a pre-trained human acupoint localization network model, and using the human acupoint localization network model to annotate the target human body data with human acupoints to obtain human acupoint annotation results; and locating human meridian acupoints of the target object based on the human acupoint annotation results and the direction of human meridians; wherein the human acupoint localization network model is trained based on the following steps: constructing a first parameterized human body model in a standard pose. The first parameterized human body model is then refined into a mesh to generate a standard 3D human body model, wherein the number of surface points in the standard 3D human body model is greater than the number of surface points in the parameterized human body model. Acupoints are marked on the vertices of the standard 3D human body model to construct a standard 3D human acupoint model. Global and / or local parameterized human body data from multiple single-viewpoints are acquired, and 3D human acupoint models corresponding to the global and / or local parameterized human body data from each single-viewpoint are generated based on the standard 3D human acupoint model. The neural network model is trained based on the global and / or local parameterized human body data from multiple single-viewpoints and the 3D human acupoint models corresponding to the global and / or local parameterized human body data from each single-viewpoint to obtain a human acupoint localization network model.

[0097] 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.

[0098] 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.

[0099] 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 method for locating acupoints along human meridians, characterized in that, include: Collect single-view global and / or local target human body data of the target object; The target human body data is input into a pre-trained human acupoint location network model, and the human acupoint location network model is used to annotate the target human body data to obtain the human acupoint annotation results. Based on the human acupoint annotation results and the direction of human meridians, the target object is located by locating human meridian acupoints, including: connecting two adjacent acupoints based on the human acupoint annotation results and the direction of human meridians to obtain multiple line segments between acupoints; locating multiple positioning points on each acupoint line segment, and finding the nearest neighbor of each positioning point on the target human body data using the nearest neighbor search method; setting a corresponding color for the nearest neighbor of each positioning point based on different meridians to obtain the human meridian acupoint location corresponding to the target object; The human acupoint location network model is trained based on the following steps: A first parametric human body model with a standard pose is constructed, and the mesh of the first parametric human body model is refined to generate a standard three-dimensional human body model. The number of surface points in the standard three-dimensional human body model is greater than the number of surface points in the first parametric human body model. Acupoints are marked on the vertices of the body surface of the standard three-dimensional human body model to construct a standard three-dimensional human acupoint model; Construct a second parametric human body model with different postures and body shapes; refine the second parametric human body model with the same mesh refinement standard as the standard three-dimensional human acupoint model to generate a three-dimensional human body model corresponding to the second parametric human body model; annotate the three-dimensional human body model with acupoint annotation data in the standard three-dimensional human acupoint model to obtain a three-dimensional human acupoint model corresponding to the three-dimensional human body model; randomly set the position of a virtual camera, and acquire global and / or local human body data of the three-dimensional human body model from multiple single perspectives through the virtual camera, as well as the three-dimensional human acupoint model corresponding to the global and / or local human body data from each single perspective; The neural network model is trained based on the global and / or local parameterized human body data under multiple single-view perspectives and the three-dimensional human acupoint model corresponding to the global and / or local parameterized human body data under each single-view perspective, to obtain the human acupoint positioning network model.

2. The method according to claim 1, characterized in that, The neural network model is trained based on the global and / or local parameterized human body data from multiple single-viewpoints and the three-dimensional human acupoint model corresponding to the global and / or local parameterized human body data from each single-viewpoint, to obtain a human acupoint localization network model, including: The global and / or local parameterized human body data under multiple single-view perspectives and the three-dimensional human acupoint model corresponding to each global and / or local parameterized human body data under each single-view perspective are input into the neural network model. The global and / or local parameterized human body data under multiple single-view perspectives are used as the model input data, and the three-dimensional human acupoint model corresponding to each global and / or local parameterized human body data under each single-view perspective is used as the ground truth to train the neural network model. During the training of the neural network model, the network parameters are adjusted using an error backpropagation algorithm. When the neural network model is trained to the convergence state, it is determined that the training of the neural network model is complete, and the human acupoint positioning network model is obtained.

3. The method according to claim 1 or 2, characterized in that, The step of marking acupoints on the vertices of the standard three-dimensional human body model to construct a standard three-dimensional human acupoint model includes: Based on the prescribed standard human acupoint location method and acupoint names, acupoints are marked on the vertices of the standard three-dimensional human body model to construct a standard three-dimensional human acupoint model.

4. The method according to claim 1, characterized in that, The method further includes: The location results of the human meridians and acupoints corresponding to the target object are displayed on an external monitor.

5. The method according to claim 1, characterized in that, The acquisition of single-view global and / or local target human body data of the target object includes: The target object's global and / or local human body data is acquired from a single perspective using the camera built into the mixed reality device.

6. A device for locating acupoints along human meridians, characterized in that, include: The acquisition module is used to acquire single-view global and / or local human body data of the target object; The annotation module is used to input the target human body data into a pre-trained human acupoint location network model, and to annotate the target human body data with human acupoints through the human acupoint location network model to obtain human acupoint annotation results. The positioning module is used to locate the acupoints of the target object based on the human acupoint annotation results and the direction of human meridians. This includes: connecting two adjacent acupoints based on the human acupoint annotation results and the direction of human meridians to obtain multiple line segments between acupoints; locating multiple positioning points on each acupoint line segment, and finding the nearest neighbor of each positioning point in the target human body data using the nearest neighbor search method; and assigning a corresponding color to the nearest neighbor of each positioning point based on different meridians to obtain the human acupoint positioning of the target object. The model training module is used to construct a first parametric human body model in a standard pose, and to refine the mesh of the first parametric human body model to generate a standard 3D human body model. The number of surface points in the standard 3D human body model is greater than that in the first parametric human body model. Acupoints are marked on the vertices of the standard 3D human body model to construct a standard 3D human acupoint model. Second parametric human body models with different poses and body types are constructed. The second parametric human body model is then refined using the same mesh as the standard 3D human acupoint model to generate a corresponding 3D human body model. The model is then trained using the standard 3D human body model. The acupoint annotation data in the three-dimensional human acupoint model is used to annotate the acupoints in the three-dimensional human acupoint model to obtain the three-dimensional human acupoint model corresponding to the three-dimensional human acupoint model; the position of the virtual camera is randomly set, and the global and / or local human data of the three-dimensional human acupoint model under multiple single views are acquired through the virtual camera, as well as the three-dimensional human acupoint model corresponding to the global and / or local human data under each single view; the neural network model is trained based on the global and / or local parameterized human data under multiple single views and the three-dimensional human acupoint model corresponding to the global and / or local parameterized human data under each single view to obtain the human acupoint localization network model.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the human meridian acupoint location method as described in any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the human meridian acupoint location method as described in any one of claims 1 to 5.

Images