Acupoint detection method, device, equipment and storage medium

Abstract

Embodiments of the present disclosure provide an acupoint detection method, device and equipment and a storage medium. The method comprises: obtaining a training data set, wherein a sample in the training data set comprises a human body point cloud model, and an acupoint position corresponding to the human body point cloud model; training a preset acupoint detection model according to the training data set to obtain an acupoint detection model with strong detection capability, and then rapidly and accurately detecting the acupoint position based on the model, thereby effectively improving the acupoint detection effect.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of computer, and particularly relates to an acupoint detection method and device, equipment and a storage medium. BACKGROUND

[0002] Acupuncture is one of the important components of traditional Chinese medicine. With the increasing demand of the public for acupuncture, there is a large gap between professional acupuncturists. In order to solve the contradiction between the number of doctors and the demand of patients, researchers focus on acupuncture robots. Acupuncture robots need to solve several problems: intelligent acupoint selection, intelligent acupoint searching, automatic needle insertion, etc., among which intelligent acupoint searching is an important part of automatic acupuncture of acupuncture robots. The current intelligent acupoint searching scheme generally has a large error and poor effect. Therefore, how to improve the acupoint detection effect has become a technical problem to be solved at present. SUMMARY

[0003] Embodiments of the present disclosure provide an acupoint detection method, device, equipment and storage medium.

[0004] In a first aspect, embodiments of the present disclosure provide a training method of an acupoint detection model, which comprises:

[0005] obtaining a training data set, wherein the samples in the training data set comprise a human body point cloud model, and the acupoint positions corresponding to the human body point cloud model;

[0006] sorting each point in the human body point cloud model to obtain the order of each point;

[0007] generating a point set corresponding to each point according to each point and its adjacent points, and generating a graph corresponding to the point set, wherein the graph is represented by a point feature matrix and a point adjacency matrix;

[0008] using a feature extraction network in a preset acupoint detection model to sequentially perform feature extraction on the point feature matrix of each graph according to the order of each point, to generate a new point feature matrix of each graph; calculating a new point adjacency matrix corresponding to the new point feature matrix of each graph according to the adjacency matrix of each graph, adjusting each new point feature matrix according to the new point adjacency matrix corresponding to each new point feature matrix, and establishing a new graph corresponding to each point according to the new point adjacency matrix corresponding to each point and the adjusted point feature matrix corresponding to each point; continuously establishing the new graph corresponding to each point until a preset number of times is reached; inputting the point feature matrix of the latest graph corresponding to each point when the preset number of times is reached into the PointNET in the preset acupoint detection model, outputting the acupoint positions, and updating the parameters of the preset acupoint detection model according to the output acupoint positions and the acupoint positions corresponding to the human body point cloud model, continuously iterating until a preset stop condition is met, to obtain a trained acupoint detection model.

[0009] In some possible implementation manners of the first aspect, the sorting of the points in the human body point cloud model obtains an order of the points, including:

[0010] The points are sorted according to the numerical values of the target features of the points, to obtain the order of the points; or,

[0011] The average value of the feature numerical values of the points is calculated, and the points are sorted according to the average value of the feature numerical values of the points, to obtain the order of the points.

[0012] In some possible implementation manners of the first aspect, the adjacent points of the points are determined through the following steps:

[0013] According to the features of the points, the adjacency degrees of the points to other points except themselves are calculated;

[0014] For any point, other points with an adjacency degree greater than or equal to a preset threshold are taken as the adjacent points of the current point; or,

[0015] The other points are sorted according to the adjacency degrees from large to small, and the first N other points are selected as the adjacent points of the current point; or,

[0016] The other points are sorted according to the adjacency degrees from small to large, and the last N other points are selected as the adjacent points of the current point.

[0017] In some possible implementation manners of the first aspect, the feature extraction network includes a convolutional neural network and a fully connected neural network;

[0018] The feature extraction network in the preset acupoint detection model is used to perform feature extraction on the point feature matrix of each graph corresponding to each point in sequence according to the order of the points, to generate a new point feature matrix of each graph, including:

[0019] The point feature matrix of each graph corresponding to each point is sequentially subjected to convolution processing by the convolutional neural network according to the order of the points, to obtain a sub-feature corresponding to each graph;

[0020] The sub-feature corresponding to each graph is input into the fully connected neural network, to obtain the new point feature matrix of each graph.

[0021] In some possible implementation manners of the first aspect, a new point adjacency matrix corresponding to the new point feature matrix of each graph is calculated according to an adjacency matrix of each graph, including:

[0022] The product of the adjacency matrix corresponding to the graph and a feature extraction operation matrix corresponding to the graph is calculated, to obtain the new point adjacency matrix corresponding to the new point feature matrix of the graph.

[0023] In a second aspect, an embodiment of the present disclosure provides an acupoint detection method, including:

[0024] Sort each point in the human point cloud model to be detected to obtain an order of the points;

[0025] According to the points and their adjacent points, a point set corresponding to each point is generated, and a graph corresponding to the point set is generated, wherein the graph is represented by a point feature matrix and a point adjacency matrix;

[0026] According to the order of the points, the point feature matrix of the graph corresponding to each point is sequentially subjected to feature extraction by using a feature extraction network in the acupoint detection model, to generate a new point feature matrix of each graph; a new point adjacency matrix corresponding to the new point feature matrix of each graph is calculated according to the adjacency matrix of each graph, each new point feature matrix is adjusted according to the new point adjacency matrix corresponding to the new point feature matrix, and a new graph corresponding to each point is established according to the new point adjacency matrix corresponding to each point and the adjusted point feature matrix corresponding to each point; the new graph corresponding to each point is continuously established until a preset number of times is reached; and the point feature matrix of the latest graph corresponding to each point when the preset number of times is reached is input into the Point NET in the acupoint detection model, and an acupoint position is output.

[0027] The acupoint detection model is obtained based on the training method of the acupoint detection model as described above.

[0028] In a third aspect, embodiments of the present disclosure provide a device for training an acupoint detection model, the device comprising:

[0029] The acquisition module is configured to acquire a training data set, wherein a sample in the training data set includes a human point cloud model and an acupoint position corresponding to the human point cloud model.

[0030] The sorting module is configured to sort each point in the human point cloud model to obtain an order of the points.

[0031] The generation module is configured to generate, according to the points and their adjacent points, a point set corresponding to each point, and generate a graph corresponding to the point set, wherein the graph is represented by a point feature matrix and a point adjacency matrix.

[0032] The training module is configured to: utilize a feature extraction network in the preset acupoint detection model to sequentially perform feature extraction on point feature matrices of graphs corresponding to the points in the order of the points, to generate new point feature matrices of the graphs; calculate new point adjacency matrices corresponding to the new point feature matrices of the graphs according to adjacency matrices of the graphs, adjust the new point feature matrices according to the new point adjacency matrices corresponding to the new point feature matrices, and establish new graphs corresponding to the points according to the new point adjacency matrices corresponding to the points and the adjusted point feature matrices corresponding to the points; continuously establish the new graphs corresponding to the points until a preset number of times is reached; input point feature matrices of the latest graphs corresponding to the points when the preset number of times is reached into the Point NET in the preset acupoint detection model, output acupoint positions, and update parameters of the preset acupoint detection model according to the output acupoint positions and acupoint positions corresponding to the human body point cloud model, continuously iterate until a preset stop condition is met, and obtain a trained acupoint detection model.

[0033] In a fourth aspect, an embodiment of the present disclosure provides an acupoint detection device, which comprises:

[0034] The sorting module is configured to sort the points in the human body point cloud model to be detected to obtain the order of the points.

[0035] The generating module is configured to generate a point set corresponding to each point according to the point and adjacent points of the point, and generate a graph corresponding to the point set, wherein the graph is represented by a point feature matrix and a point adjacency matrix.

[0036] The detection module is configured to: utilize a feature extraction network in the acupoint detection model to sequentially perform feature extraction on point feature matrices of graphs corresponding to the points in the order of the points, to generate new point feature matrices of the graphs; calculate new point adjacency matrices corresponding to the new point feature matrices of the graphs according to adjacency matrices of the graphs, adjust the new point feature matrices according to the new point adjacency matrices corresponding to the new point feature matrices, and establish new graphs corresponding to the points according to the new point adjacency matrices corresponding to the points and the adjusted point feature matrices corresponding to the points; continuously establish the new graphs corresponding to the points until a preset number of times is reached; and input point feature matrices of the latest graphs corresponding to the points when the preset number of times is reached into the Point NET in the acupoint detection model, to output acupoint positions.

[0037] The acupoint detection model is obtained based on the training method of the acupoint detection model.

[0038] In a fifth aspect, an embodiment of the present disclosure provides an electronic device, which comprises: at least one processor; and a memory connected with the at least one processor in communication; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method as described above.

[0039] In a sixth aspect, the embodiments of the present disclosure provide a non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the method described above.

[0040] In the embodiments of the present disclosure, a training data set can be acquired, wherein a sample in the training data set includes a human body point cloud model and an acupoint position corresponding to the human body point cloud model, a preset acupoint detection model is trained according to the training data set, and an acupoint detection model with strong detection capability is obtained, so that the acupoint position is quickly and accurately detected based on the model, and the acupoint detection effect is effectively improved.

[0041] It should be understood that the content described in the summary section is not intended to limit the key or important features of the embodiments of the present disclosure, nor to limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0042] The above and other features, advantages, and aspects of the embodiments of the present disclosure will become more apparent as various embodiments of the present disclosure are described in detail with reference to the drawings. The drawings are intended to better understand the present disclosure and do not limit the present disclosure. In the drawings, the same or similar reference numerals refer to the same or similar elements, and:

[0043] Figure 1 A flowchart of an acupoint detection model training method provided by the embodiments of the present disclosure is shown;

[0044] Figure 2 A schematic diagram of a human body point cloud model is shown;

[0045] Figure 3 A combined projection schematic diagram of an adjusted point feature matrix corresponding to each point is shown;

[0046] Figure 4 A flowchart of an acupoint detection method provided by the embodiments of the present disclosure is shown;

[0047] Figure 5 An acupoint detection effect comparison schematic diagram is shown;

[0048] Figure 6 A structural diagram of an acupoint detection model training device provided by the embodiments of the present disclosure is shown;

[0049] Figure 7 A structural diagram of an acupoint detection device provided by the embodiments of the present disclosure is shown;

[0050] Figure 8 A structural diagram of an exemplary electronic device capable of implementing the embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0051] In order to make the purposes, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative work belong to the protection scope of the present disclosure.

[0052] In addition, the term "and / or" in this paper is only to describe the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the three cases of A alone, A and B together, and B alone. In addition, the character " / " in this paper generally represents an "or" relationship between the front and rear associated objects.

[0053] In view of the problems in the background art, the embodiments of the present disclosure provide an acupoint detection method, device, equipment and storage medium. Specifically, a training data set is obtained, wherein the samples in the training data set include a human body point cloud model and an acupoint position corresponding to the human body point cloud model, a preset acupoint detection model is trained according to the training data set to obtain an acupoint detection model with strong detection capability, and then the acupoint position is quickly and accurately detected based on the model, thereby effectively improving the acupoint detection effect.

[0054] The acupoint detection method, device, equipment and storage medium provided by the embodiments of the present disclosure will be described in detail below with reference to the drawings and specific embodiments.

[0055] Figure 1 A flowchart of an acupoint detection model training method provided by the embodiments of the present disclosure is shown in FIG. 1. Figure 1 As shown in FIG. 1, the training method 100 can include the following steps:

[0056] S110, obtaining a training data set.

[0057] The samples in the training data set include a human body point cloud model and an acupoint position corresponding to the human body point cloud model. For example, as shown in FIG. 2, a certain human body point cloud model can be obtained. Figure 2 As shown in FIG. 2, the acupoint position can be the position of Dazhui acupoint, Fengmen acupoint, Shendao acupoint, etc., which is not limited here.

[0058] S120, sorting each point in the human body point cloud model to obtain the order of each point.

[0059] Optionally, each point can be sorted according to the numerical value of the target feature of each point to quickly obtain the order of each point.

[0060] For example, the points can be sorted according to the X-axis coordinate size, the Y-axis coordinate size, or the Z-axis coordinate size to obtain the order of the points.

[0061] In addition, the average value of the feature values of the points can be calculated, and the points can be sorted according to the average value of the feature values of the points to quickly obtain the order of the points.

[0062] For example, the average value of the X-axis coordinate, the Y-axis coordinate, and the Z-axis coordinate of the points can be calculated, and the points can be sorted according to the average value of the coordinates of the points to obtain the order of the points.

[0063] In S130, a point set corresponding to each point is generated according to the point and the adjacent points of the point, and a graph corresponding to the point set is generated.

[0064] The point set includes the point and the adjacent points of the point, and the point is the center point of the point set; the graph is jointly represented by a point feature matrix and a point adjacency matrix, the point feature matrix is used to represent the features of the points in the point set, and the point adjacency matrix is used to represent the adjacency relationship between the points in the point set and the adjacent points of the points.

[0065] It is worth noting that the adjacent points of each point can be quickly determined by the following steps:

[0066] According to the features of the points, the adjacency degrees of the points and other points except themselves are calculated.

[0067] For example, the Euclidean distance, the shortest path length, or the correlation between the points and other points except themselves can be calculated, and the Euclidean distance, the shortest path length, or the correlation between the points and other points except themselves can be taken as the adjacency degrees of the points and other points except themselves.

[0068] For any point, other points with an adjacency degree greater than or equal to a preset threshold value can be taken as the adjacent points of the current point, wherein the preset threshold value can be flexibly set as needed, and is not limited herein.

[0069] Alternatively, other points can be sorted according to the adjacency degrees from large to small, and the first N other points can be selected as the adjacent points of the current point.

[0070] Alternatively, other points can be sorted according to the adjacency degrees from small to large, and the last N other points can be selected as the adjacent points of the current point.

[0071] In S140, a feature extraction network in the preset acupoint detection model is used to sequentially perform feature extraction on the point feature matrix of each graph according to the order of the points, to generate a new point feature matrix of each graph; a new point adjacency matrix corresponding to the new point feature matrix of each graph is calculated according to the adjacency matrix of each graph, each new point feature matrix is adjusted according to the new point adjacency matrix corresponding to the new point feature matrix, and a new graph corresponding to each point is established according to the new point adjacency matrix corresponding to each point and the adjusted point feature matrix corresponding to each point; the new graph corresponding to each point is continuously established until a preset number of times is reached; the point feature matrix of the latest graph corresponding to each point when the preset number of times is reached is input into the Point NET in the preset acupoint detection model, an acupoint position is output, and the parameters of the preset acupoint detection model are updated according to the output acupoint position and the acupoint position corresponding to the human body point cloud model, and the updating is continuously iterated until a preset stop condition is met, to obtain a trained acupoint detection model.

[0072] In some embodiments, the feature extraction network can include a convolutional neural network and a fully connected neural network. Accordingly, the point feature matrix of each graph can be sequentially processed by the convolutional neural network according to the order of the points to obtain sub-features corresponding to each graph, and then the sub-features corresponding to each graph are input into the fully connected neural network, to quickly obtain the new point feature matrix of each graph.

[0073] In some embodiments, the product of the adjacency matrix corresponding to the graph and the feature extraction operation matrix corresponding to the graph can be calculated and used as the new point adjacency matrix corresponding to the new point feature matrix of the graph. The feature extraction operation matrix is used to represent the operation process of the point feature matrix during feature extraction.

[0074] It can be understood that the continuously establishing the new graph corresponding to each point in S140 until the preset number of times is reached means continuously updating the graph corresponding to each point until the preset number of times is reached. The preset stop condition can be that the model loss value is less than a preset threshold, or that the number of iterations reaches a specified number.

[0075] For example, the combined projection of the adjusted point feature matrix corresponding to each point under different convolution kernels can be as shown in Figure 3 .

[0076] In the embodiments of the present disclosure, the human body point cloud model can be converted into a graph, the local features of the human body point cloud model can be obtained through the graph, which can effectively supplement the problem that the PointNet network cannot extract local features, and at the same time, the feature extraction efficiency can be improved by sorting the points in the human body point cloud model. On this basis, an acupoint detection model with strong detection capability can be quickly trained.

[0077] Based on the training method 100 of the acupoint detection model provided in the embodiments of the present disclosure, the present disclosure further provides an acupoint detection method, as shown inFigure 4 As shown, the acupoint detection method 400 may include the following steps:

[0078] S410 sorts the points in the human point cloud model to be detected to obtain the order of the points.

[0079] In some embodiments, a 3D scanning device (e.g., a 3D scanner) can be used to scan the human body to be detected, obtain point cloud data of the human body to be detected, and quickly construct a point cloud model of the human body to be detected based on the point cloud data of the human body to be detected.

[0080] S420: Generate a set of points corresponding to each point and its adjacent points, and generate a graph corresponding to the set of points. The graph is represented by a point feature matrix and a point adjacency matrix.

[0081] S430: Using the feature extraction network in the acupoint detection model, features are extracted from the point feature matrix of the graph corresponding to each point in sequence, generating new point feature matrices for each graph; the new point adjacency matrix corresponding to the new point feature matrix of each graph is calculated based on the adjacency matrix of each graph; the new point feature matrix is ​​adjusted based on the new point adjacency matrix corresponding to each new point feature matrix; a new graph corresponding to each point is established based on the new point adjacency matrix and the corresponding adjusted point feature matrix; new graphs corresponding to each point are continuously established until a preset number of times is reached; the point feature matrix of the latest graph corresponding to each point when the preset number of times is reached is input into the Point NET in the acupoint detection model, and the acupoint location is output.

[0082] The acupoint detection model is obtained based on the training method of the acupoint detection model described above.

[0083] In the embodiments of this disclosure, acupoint locations can be detected quickly and accurately based on acupoint detection models with strong detection capabilities, effectively improving the acupoint detection effect.

[0084] The following can be combined Figure 5 The effects of the acupoint detection method provided in the embodiments of this disclosure are intuitively demonstrated, such as... Figure 5 As shown, the point pointed to by the label is the actual location of the Dazhui acupoint, the point pointed to by AOP is the location of the Dazhui acupoint detected by the acupoint detection method provided in the embodiments of this disclosure, the point pointed to by Edge Conv is the location of the Dazhui acupoint detected by the acupoint detection method based on Edge Conv, the point pointed to by PointNet is the location of the Dazhui acupoint detected by the acupoint detection method based on PointNet, and the point pointed to by PointNet++ is the location of the Dazhui acupoint detected by the acupoint detection method based on PointNet++. It can be seen that the point pointed to by AOP is closest to the point pointed to by the label, meaning that the detection effect of the acupoint detection method provided in the embodiments of this disclosure is significantly better than the detection effect of existing acupoint detection methods.

[0085] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a combination of a series of actions, but those skilled in the art should know that the disclosure is not limited by the order of the described actions, because according to the disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the disclosure.

[0086] The above is the introduction of the method embodiment, and the scheme of the disclosure will be further described through the device embodiment.

[0087] Figure 6 The structure diagram of the training device of an acupoint detection model provided by an embodiment of the disclosure is shown in FIG. 6. Figure 6 As shown in FIG. 6, the training device 600 can include:

[0088] The acquisition module 610 is configured to acquire a training data set, wherein a sample in the training data set includes a human body point cloud model and an acupoint position corresponding to the human body point cloud model.

[0089] The sorting module 620 is configured to sort each point in the human body point cloud model to obtain an order of each point.

[0090] The generation module 630 is configured to generate a point set corresponding to each point according to each point and its adjacent points, and generate a graph corresponding to the point set, wherein the graph is represented by a point feature matrix and a point adjacency matrix.

[0091] The training module 640 is configured to use a feature extraction network in a preset acupoint detection model to sequentially perform feature extraction on the point feature matrix of each graph according to the order of each point, to generate a new point feature matrix of each graph; calculate a new point adjacency matrix corresponding to the new point feature matrix of each graph according to the adjacency matrix of each graph, adjust each new point feature matrix according to the new point adjacency matrix corresponding to the new point feature matrix, and establish a new graph corresponding to each point according to the new point adjacency matrix corresponding to each point and the adjusted point feature matrix corresponding to each point; constantly establish the new graph corresponding to each point until a preset number of times are reached; input the point feature matrix of the latest graph corresponding to each point when the preset number of times are reached into the Point NET in the preset acupoint detection model, output an acupoint position, and update the parameters of the preset acupoint detection model according to the output acupoint position and the acupoint position corresponding to the human body point cloud model, constantly iterate until a preset stop condition is met, and obtain a trained acupoint detection model.

[0092] In some embodiments, the sorting module 620 is specifically configured to:

[0093] The points are sorted according to the numerical values of the target features of the points, to obtain an order of the points; or

[0094] The average value of the feature numerical values of the points is calculated, and the points are sorted according to the average value of the feature numerical values of the points, to obtain an order of the points.

[0095] In some embodiments, the adjacent points of each point are determined by the following steps:

[0096] According to the features of the points, the adjacency degrees of each point to other points except itself are calculated;

[0097] For any point, other points with an adjacency degree greater than or equal to a preset threshold are taken as the adjacent points of the current point; or

[0098] The other points are sorted according to the adjacency degrees from large to small, and the first N other points are selected as the adjacent points of the current point; or

[0099] The other points are sorted according to the adjacency degrees from small to large, and the last N other points are selected as the adjacent points of the current point.

[0100] In some embodiments, the feature extraction network includes a convolutional neural network and a fully connected neural network.

[0101] The training module 640 is specifically configured to:

[0102] The point feature matrices of the graphs corresponding to the points are sequentially processed by the convolutional neural network according to the order of the points, to obtain sub-features corresponding to the graphs;

[0103] The sub-features corresponding to the graphs are input into the fully connected neural network, to obtain new point feature matrices of the graphs.

[0104] In some embodiments, the training module 640 is specifically configured to:

[0105] The product of the adjacency matrix corresponding to the graph and the feature extraction operation matrix corresponding to the graph is calculated, to obtain a new point adjacency matrix corresponding to the new point feature matrix of the graph.

[0106] It can be understood that, Figure 6 Each module / unit in the training device 600 shown has the function of implementing each step in the training method 100 shown and can achieve the corresponding technical effects thereof, and for the sake of brevity, will not be repeated here. Figure 1 Each step in the training method 100 shown has the function of implementing each step in the training method 100 shown and can achieve the corresponding technical effects thereof, and for the sake of brevity, will not be repeated here.

[0107] Figure 7 A structural diagram of an acupoint detection device provided by an embodiment of the present disclosure is shown, as shown in Figure 7 The acupoint detection device 700 can include:

[0108] The sorting module 710 is configured to sort each point in the human point cloud model to be detected to obtain an order of the points.

[0109] The generating module 720 is configured to generate a point set corresponding to each point according to the point and the adjacent points of the point, and generate a graph corresponding to the point set, where the graph is represented by a point feature matrix and a point adjacency matrix.

[0110] The detecting module 730 is configured to use a feature extraction network in the acupoint detection model to perform feature extraction on the point feature matrix of each graph in the order of the points, to generate a new point feature matrix of each graph; calculate a new point adjacency matrix corresponding to the new point feature matrix of each graph according to the adjacency matrix of each graph, adjust each new point feature matrix according to the new point adjacency matrix corresponding to the new point feature matrix, and establish a new graph corresponding to each point according to the new point adjacency matrix corresponding to the point and the adjusted point feature matrix corresponding to the point; constantly establish the new graph corresponding to each point until a preset number of times is reached; and input the point feature matrix of the latest graph corresponding to each point when the preset number of times is reached into the Point NET in the acupoint detection model, and output an acupoint position.

[0111] The acupoint detection model is obtained based on the training method described above.

[0112] It can be understood that, Figure 7 The modules / units in the acupoint detection device 700 shown have the functions of implementing each step in the acupoint detection method 400 shown, and can achieve the corresponding technical effects. For the sake of brevity, they will not be described here. Figure 4 The modules / units in the acupoint detection device 700 shown have the functions of implementing each step in the acupoint detection method 400 shown, and can achieve the corresponding technical effects. For the sake of brevity, they will not be described here.

[0113] Figure 8 A block diagram illustrating an example electronic device in which embodiments of the present disclosure can be implemented is shown. The electronic device 800 is intended to represent various forms of digital computers, such as laptops, desktops, tablets, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device 800 can also represent various forms of mobile devices, such as personal digital assistants, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown here, their connections and relationships, and their functions, are meant to be examples only, and are not intended to limit the implementations of the present disclosure described and / or claimed in this document.

[0114] As Figure 8As shown, the electronic device 800 can include a computing unit 801 that can perform various appropriate actions and processes in accordance with a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. Various programs and data required for the operation of the electronic device 800 can also be stored in the RAM 803. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other through a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0115] A plurality of components in the electronic device 800 are connected to the I / O interface 805, including an input unit 806 such as a keyboard, a mouse, and the like, an output unit 807 such as various types of displays, a speaker, and the like, a storage unit 808 such as a magnetic disk, an optical disk, and the like, and a communication unit 809 such as a network card, a modem, a wireless communication transceiver, and the like. The communication unit 809 allows the electronic device 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0116] The computing unit 801 can be various general and / or special-purpose processing components having processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, and the like. The computing unit 801 performs various methods and processes described above, such as the method 100 or the method 400. For example, in some embodiments, the method 100 or the method 400 can be implemented as a computer program product including a computer program tangibly embodied in a computer-readable medium, such as the storage unit 808. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the method 100 or the method 400 described above can be performed. Alternatively, in other embodiments, the computing unit 801 can be configured to perform the method 100 or the method 400 by any other appropriate means, such as by means of firmware.

[0117] The various implementations described above can be implemented in digital electronic circuitry, integrated circuitry, specially designed application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0118] Program code for carrying out methods of the present disclosure can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program code, when executed by the processor or controller, produces a means for implementing the functions / acts specified in the flowcharts and / or block diagrams. The program code can be executed entirely on a machine, partially on a machine, partially on a machine as a stand-alone software package, partially on a machine and partially on a remote machine or entirely on a remote machine or server.

[0119] In the context of the present disclosure, a computer-readable medium can be a tangible medium that can contain or store program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the computer-readable storage medium can include one or more lines of electrical connections, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0120] It should be noted that the present disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to make a computer execute the method 100 or the method 400, and achieve the corresponding technical effects of the embodiments of the present disclosure executing their methods. For brevity, the description will not be repeated here.

[0121] In addition, the present disclosure also provides a computer program product comprising a computer program which, when executed by a processor, implements the method 100 or the method 400.

[0122] To provide for interaction with a user, the above described embodiments can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0123] The above described embodiments can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0124] The computer system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server is generally established by computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0125] It should be understood that the various forms of flow shown above can be re-ordered, added to, or deleted from without departing from the scope of the present disclosure. For example, the steps recited in the present disclosure can be performed in parallel, in series, or in a different order, without departing from the desired results of the technical solutions disclosed in the present disclosure, and are not limited herein.

[0126] The above detailed description does not limit the scope of the disclosure. Various modifications, combinations, sub-combinations and alternatives can be made to the detailed description. Any modification, equivalent replacement and improvement etc. made within the spirit and principle of the disclosure shall be included in the scope of the disclosure.

Claims

1. A method for training an acupoint detection model, characterized in that, The method comprises: obtaining a training data set, wherein the samples in the training data set comprise human point cloud models, and the human point cloud models correspond to acupoint positions; sorting each point in the human point cloud model to obtain the order of each point; generating a point set corresponding to each point according to each point and its adjacent points, and generating a graph corresponding to the point set, wherein the graph is represented by a point feature matrix and a point adjacency matrix; using a feature extraction network in a preset acupoint detection model to sequentially extract features of the point feature matrix of each graph according to the order of each point, to generate a new point feature matrix of each graph; calculating a new point adjacency matrix corresponding to the new point feature matrix of each graph according to the adjacency matrix of each graph, adjusting each new point feature matrix according to the new point adjacency matrix corresponding to the new point feature matrix, and establishing a new graph corresponding to each point according to the new point adjacency matrix and the adjusted point feature matrix corresponding to each point; continuously establishing a new graph corresponding to each point until a preset number of times is reached; inputting the point feature matrix of the latest graph corresponding to each point when the preset number of times is reached into the Point NET in the preset acupoint detection model, outputting an acupoint position, and updating the parameters of the preset acupoint detection model according to the output acupoint position and the acupoint position corresponding to the human point cloud model, continuously iterating until a preset stopping condition is met, and obtaining a trained acupoint detection model; the sorting each point in the human point cloud model to obtain the order of each point comprises: sorting each point according to the numerical value of the target feature of each point to obtain the order of each point; or calculating the average value of the feature numerical value of each point, and sorting each point according to the size of the average value of the feature numerical value of each point to obtain the order of each point; the adjacent points of each point are determined by the following steps: calculating the adjacency degree of each point with other points except itself according to the feature of each point; for any point, other points with an adjacency degree greater than or equal to a preset threshold are taken as the adjacent points of the current point; or sorting other points according to the adjacency degree from large to small, and selecting the first N other points as the adjacent points of the current point; or sorting other points according to the adjacency degree from small to large, and selecting the last N other points as the adjacent points of the current point.

2. The method of claim 1, wherein, The feature extraction network comprises a convolutional neural network and a fully connected neural network; the using a feature extraction network in a preset acupoint detection model to sequentially extract features of the point feature matrix of each graph according to the order of each point to generate a new point feature matrix of each graph comprises: convolving the point feature matrix of each graph corresponding to each point by the convolutional neural network in the order of each point to obtain a sub-feature corresponding to each graph; inputting the sub-feature corresponding to each graph into the fully connected neural network to obtain a new point feature matrix of each graph.

3. The method of claim 1, wherein, The calculating a new point adjacency matrix corresponding to the new point feature matrix of each graph according to the adjacency matrix of each graph comprises: calculating the product of the adjacency matrix corresponding to the graph and the feature extraction operation matrix corresponding to the graph to obtain a new point adjacency matrix corresponding to the new point feature matrix of the graph.

4. An acupoint detection method, characterized by, The method comprises: sorting each point in the human point cloud model to obtain the order of each point; According to each point and its adjacent points, a point set corresponding to each point is generated, and a graph corresponding to the point set is generated, wherein the graph is represented by a point feature matrix and a point adjacency matrix; According to the order of each point, the feature extraction network in the acupoint detection model is used to sequentially perform feature extraction on the point feature matrix of each graph, to generate a new point feature matrix of each graph; a new point adjacency matrix corresponding to the new point feature matrix of each graph is calculated according to the adjacency matrix of each graph, each new point feature matrix is adjusted according to the new point adjacency matrix corresponding to the new point feature matrix, and a new graph corresponding to each point is established according to the new point adjacency matrix and the adjusted point feature matrix corresponding to each point; the new graph corresponding to each point is continuously established until a preset number of times is reached; and the point feature matrix of the latest graph corresponding to each point when the preset number of times is reached is input into the Point NET in the acupoint detection model, and an acupoint position is output. The acupoint detection model is obtained based on the training method of the acupoint detection model in any one of claims 1-3. 5.A device for training an acupoint detection model, characterized in that, The device comprises: An acquisition module configured to acquire a training data set, wherein a sample in the training data set comprises a human body point cloud model and an acupoint position corresponding to the human body point cloud model; An ordering module configured to order each point in the human body point cloud model to obtain an order of each point; A generation module configured to generate a point set corresponding to each point according to each point and its adjacent points, and generate a graph corresponding to the point set, wherein the graph is represented by a point feature matrix and a point adjacency matrix; A training module configured to use a feature extraction network in a preset acupoint detection model to sequentially perform feature extraction on the point feature matrix of each graph according to the order of each point, to generate a new point feature matrix of each graph; calculate a new point adjacency matrix corresponding to the new point feature matrix of each graph according to the adjacency matrix of each graph, adjust each new point feature matrix according to the new point adjacency matrix corresponding to the new point feature matrix, and establish a new graph corresponding to each point according to the new point adjacency matrix and the adjusted point feature matrix corresponding to each point; continuously establish the new graph corresponding to each point until a preset number of times is reached; input the point feature matrix of the latest graph corresponding to each point when the preset number of times is reached into the Point NET in the preset acupoint detection model, output an acupoint position, and update parameters of the preset acupoint detection model according to the output acupoint position and the acupoint position corresponding to the human body point cloud model, continuously iterate until a preset stop condition is met, and obtain a trained acupoint detection model; The ordering module is specifically configured to: order each point according to the numerical value of the target feature of each point to obtain the order of each point; or calculate the average value of the feature numerical value of each point, and order each point according to the size of the average value of the feature numerical value of each point to obtain the order of each point; The adjacent points of each point are determined by the following steps: According to the features of each point, the adjacency degrees of each point to other points except itself are calculated. For any point, other points with an adjacency degree greater than or equal to a preset threshold are taken as the adjacent points of the current point; or According to the adjacency degrees from large to small, the first N other points are selected as the adjacent points of the current point; or According to the adjacency degree from small to large, other points are sorted, and the last N other points are selected as the adjacent points of the current point.

6. An acupoint detection device, characterized by comprising: The device comprises: The sorting module is configured to sort each point in the human point cloud model to be detected to obtain an order of the points. The generating module is configured to generate a point set corresponding to each point according to the point and its adjacent points, and generate a graph corresponding to the point set, wherein the graph is represented by a point feature matrix and a point adjacency matrix. The detecting module is configured to use a feature extraction network in the acupoint detection model to sequentially extract features of the point feature matrix of each graph according to the order of the points, generate a new point feature matrix of each graph, calculate a new point adjacency matrix corresponding to the new point feature matrix of each graph according to the adjacency matrix of each graph, adjust each new point feature matrix according to the new point adjacency matrix corresponding to the new point feature matrix, and establish a new graph corresponding to each point according to the new point adjacency matrix corresponding to the point and the adjusted point feature matrix corresponding to the point; the new graph corresponding to each point is continuously established until a preset number of times is reached; and the point feature matrix of the latest graph corresponding to each point when the preset number of times is reached is input into the Point NET in the acupoint detection model to output an acupoint position. The acupoint detection model is obtained based on the training method of the acupoint detection model in any one of claims 1-3.

7. An electronic device, comprising: The electronic device comprises: At least one processor; and a memory connected in communication with the at least one processor; wherein The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method of any one of claims 1-4.

8. A non-transitory computer-readable storage medium having stored thereon computer instructions, wherein, The computer instructions are used to enable the computer to execute the method of any one of claims 1-4. The computer instructions are used to enable the computer to execute the method of any one of claims 1-4.

Images