CN112364785B - Exercise training guiding method, device, equipment and computer storage medium - Google Patents

Abstract

The embodiment of the application provides a method, a device, equipment and a computer storage medium for guiding exercise training, wherein the method comprises the following steps: acquiring a video image of an athlete; identifying joint key points of the athlete in the video image, and obtaining joint key point data of the athlete during the movement; obtaining corresponding athlete muscle point force data from the joint key point data according to a muscle force adjacency matrix model; and comparing the athlete muscle point force data with the standard action muscle point force data to obtain a comparison result, and providing exercise training guidance according to the comparison result. According to the embodiment provided by the application, the muscle strength data of the athlete is obtained by identifying the joint key points of the athlete and utilizing the muscle strength adjacency matrix model according to the change of the skeletal key points, and the muscle strength data is compared with the muscle strength of standard actions to obtain the muscle strength advice and the exercise advice.

Description

A sports training guidance method, device, equipment and computer storage medium

technical field

The present application belongs to the technical field of action recognition, and in particular relates to a sports training guidance method, device, equipment and computer storage medium.

Background technique

With the development of science and technology, action recognition technology has also developed greatly, and action recognition technology has great application value in the field of auxiliary training. It can be used in fields such as sports and dance to analyze, evaluate and assist training of professional technical movements.

In the implementation of the existing technology, wearable devices such as wristbands and patches are used to fix the movement direction and speed at the extremities of the limbs, combined with the Kinect depth camera, to provide guidance by judging the gap between the joint position and movement sequence and the standard movement, and comparing it with the standard movement.

However, the existing technology is based on a simple comparison of joint position differences, which can only provide right and wrong judgment results compared with standard actions, and cannot provide suggestions and differences in force exertion, resulting in low training efficiency.

Contents of the invention

The embodiment of the present application provides a sports training guidance method, device, equipment and computer storage medium, which can solve the problems in the prior art that the differences in muscle power and guidance suggestions cannot be provided, and the training guidance efficiency is low.

In the first aspect, the embodiment of the present application provides a method for instructing sports training, the method includes:

Obtain video images of athletes;

Identify the key points of the joints of the athletes in the video image, and obtain the data of the key points of the joints of the athletes when they are in motion;

According to the muscle force adjacency matrix model, the corresponding athlete's muscle point force data is obtained from the joint key point data;

Comparing the athlete's muscle point force data with the standard action muscle point force data to obtain the comparison results, and provide sports training guidance based on the comparison results.

In one embodiment, the muscle power adjacency matrix model includes:

Obtain the position data of the key points of the standard movement joints and the corresponding force data of the muscle points;

Establishing a neural network training set based on the position data of joint key points of the standard motion action and the corresponding muscle point force data;

Establish a bipartite graph network of the corresponding relationship between joint key points and muscle points;

Optimizing the bipartite graph network of the corresponding relationship between the joint key points and the muscle points according to the neural network training set to obtain the muscle power adjacency matrix model.

In one embodiment, identifying the joint key points of the athlete in the video image, and obtaining the joint key point data of the athlete when moving, including:

Set the initial spatial coordinates of the camera, and establish a spatial coordinate system based on the initial spatial coordinates;

Identifying key points of joints of athletes in the video image, and obtaining the space coordinates of the key points based on the space coordinate system;

Establishing a first coordinate system based on arbitrary joint key points, and converting the key point space coordinates into joint key point coordinates based on the first coordinate system;

The joint key point data is obtained according to the change of the joint key point during the movement of the athlete.

In one embodiment, identifying the joint key points of the athlete in the video image, and obtaining the key point space coordinates based on the space coordinate system, includes:

Establish a space coordinate system based on the relative positions of the three cameras;

Convert each frame pixel of the video image of the athlete's joint key points into projection lines based on the three cameras respectively under the space coordinate system;

calculating mutual foot points of said projection lines;

Taking the mean value of the mutual foot points as the key point space coordinates of the joint key points based on the space coordinate system.

In one embodiment, establishing the first coordinate system based on any joint key point includes:

Select the key point of the neck joint as the origin, the left and right shoulders as the x-axis, the vertical direction as the z-axis, the y-axis located on the horizontal plane and perpendicular to the x-axis and z-axis, and establish the first coordinate system with the x, y, z-axis and the origin.

In one embodiment, the identifying the joint key points of the athlete in the video image includes:

A stacked hourglass network algorithm is used to detect each frame of the video image to identify key points of the athlete's joints.

In one embodiment, the joint key point change data during the athlete's movement is compared with the joint key point change data of the standard movement action;

Sports training guidance is provided according to the results of the position comparison of the key points of the joints.

In the second aspect, the embodiment of the present application provides a sports training guidance device, the device includes: a camera and a central processing unit;

The camera is used to obtain video images of athletes;

The central processing unit includes: a joint key point identification module, a muscle force processing module, and a sports training guidance module;

The joint key point identification module is used to identify the joint key points of the athlete in the video image, and obtain the joint key point data of the athlete during exercise;

The muscle force processing module is used to obtain corresponding athlete muscle point force data from the joint key point data according to the muscle force adjacency matrix model;

The sports training guidance module is used to compare the athlete's muscle point force data with the standard action muscle point force data to obtain a comparison result, and provide sports training guidance according to the comparison result.

In a third aspect, an embodiment of the present application provides a sports training guidance device, which includes: a camera, a processor, and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement the sports training guidance method as described above.

In a fourth aspect, an embodiment of the present application provides a computer storage medium, where computer program instructions are stored on the computer storage medium, and when the computer program instructions are executed by a processor, the above sports training guidance method is implemented.

The sports training guidance method, device, equipment, and computer storage medium provided in the embodiments of the present application can detect the movement of the athlete, extract the location information of the key points of the athlete's joints, use the muscle force adjacency matrix model to calculate the muscle force of the athlete based on the position information of the key points of the joint, compare the muscle force data with the muscle force of the standard action, and obtain the difference between the user's muscle force and the standard muscle force, and provide sports training guidance and suggestions.

Description of drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the accompanying drawings used in the embodiments of the present application will be briefly introduced below. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative work.

FIG. 1 is a schematic flow chart of a method for instructing sports training provided by an embodiment of the present application;

Fig. 2 is a bipartite network diagram of the corresponding relationship between key points and muscle points in a kind of sports training guidance method provided by the embodiment of the present application;

Fig. 3 is a schematic diagram of a method for establishing a spatial coordinate system in a sports training guidance method provided by an embodiment of the present application;

Fig. 4 is a schematic structural diagram of a stacked hourglass network algorithm in a sports training guidance method provided by an embodiment of the present application;

Fig. 5 is a schematic structural diagram of a sports training guidance device provided by an embodiment of the present application;

Fig. 6 is a schematic diagram of a hardware structure of a sports training guidance device provided by an embodiment of the present application.

Detailed ways

The characteristics and exemplary embodiments of various aspects of the application will be described in detail below. In order to make the purpose, technical solution and advantages of the application clearer, the application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described here are only intended to explain the present application rather than limit the present application. It will be apparent to one skilled in the art that the present application may be practiced without some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by showing examples of the present application.

It should be noted that in this document, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the term "comprises", "comprises" or any other variation thereof is intended to cover a non-exclusive inclusion such that a process, method, article or apparatus comprising a set of elements includes not only those elements but also other elements not expressly listed or which are inherent to such process, method, article or apparatus. Without further limitations, an element defined by the statement "comprising..." does not exclude the presence of additional same elements in the process, method, article or device comprising said element.

The embodiments provided in this application can be used in fields such as sports and dance, and can provide training guidance, analysis and evaluation for technical movements.

However, the existing technology is based on a simple comparison of joint position differences, which can only provide judgment results of right and wrong compared with standard movements, but cannot provide force differences and suggestions, resulting in low training efficiency.

In order to solve the existing technical problems, the embodiment of the present application provides a sports training guidance method, device, equipment and computer storage medium. By identifying key points of the joints in the video image of the athlete, the muscle power adjacency matrix model is used to calculate the user's muscle power according to the position information of the joint key points, and compared with the muscle power of the standard action.

Firstly, the sports training guidance method provided by the embodiment of the present application will be introduced below.

Please refer to FIG. 1 , which shows a schematic flowchart of a sports training guidance method provided by an embodiment of the present application.

In this embodiment, the following steps may be included:

S1: Acquire video images of athletes.

The technical solution provided by the embodiment of the present application is based on processing the video images captured by the camera without wearing an information extraction device; since the device is fixed at the extremity, kinematics inverse solution is required to obtain the position of the non-terminal joint, and the kinematics inverse solution can obtain multiple analysis, and the analysis becomes extremely difficult under the high degree of freedom of multiple joints, and it is impossible to predict the position of each internal joint point. At the same time, the equipment is cumbersome to wear, and the calibration of the sensing equipment requires professionals, which increases the difficulty of use. Therefore, the technical solution provided by the embodiment of the present application reduces the process of wearing and calibrating the wearable device, and improves the user experience.

S2: Identify the key points of the joints of the athlete in the video image, and obtain the data of the key points of the joints of the athlete when he is moving.

In this embodiment, the camera used may be a depth camera, and using the depth camera to capture video images can more conveniently and effectively perform operations such as information extraction, joint key point recognition, and motion tracking.

Process the video images captured by the depth camera, identify the key points of the athlete's joints, and track the movement trajectory of the key points of the joints.

S3: According to the muscle force adjacency matrix model, obtain the corresponding muscle point force data of the athlete from the joint key point data;

Since the strength of the surface EMG signal at the same part of the same movement is linearly proportional to the strength of the muscle force, the surface EMG signal can be equated with the strength of the muscle force.

In this embodiment, a standard graph neural network training data set is established, the muscle electrical signal strength of professional athletes performing related actions is collected, and the above data is used as standard action data to optimize the parameters of the adjacency matrix model.

The corresponding relationship between joint key points and muscle force can be obtained from the above-mentioned adjacency matrix model, so the muscle force data can be calculated from the position information and motion state of key points of bones (joints).

S4: Comparing the athlete's muscle point force data with the standard action muscle point force data to obtain a comparison result, and providing sports training guidance according to the comparison result.

Through the above-mentioned embodiments, it is possible to obtain the difference between the user’s muscle force and the standard muscle force; to obtain force force recommendations and muscle node exercise suggestions, which solves the problem in the prior art that only based on simple comparisons of joint position differences, it can only provide judgments about the right and wrong of actions, and cannot provide improvement suggestions for muscle force conditions, resulting in low training efficiency. This embodiment can not only provide strength suggestion guidance, but also provide action guidance suggestions based on the identified key points of the athlete's joints, and can also combine action guidance with muscle force guidance to provide professional training guidance and effectively improve training efficiency.

Please refer to Fig. 2, which is a bipartite graph of the corresponding relationship between key points and muscle points in a sports training guidance method provided by the embodiment of the present application. In this embodiment, the muscle power adjacency matrix model includes:

Obtain the position data of the key points of the standard movement joints and the corresponding force data of the muscle points;

A neural network training set is established based on the joint key point position data of the standard motion action and corresponding muscle point force data; a bipartite graph network of the corresponding relationship between the joint key point and the muscle point is established; the bipartite graph network of the joint key point and the muscle point is optimized according to the neural network training set, and the muscle force adjacency matrix model is obtained.

In the above embodiment, the key points selected (skeletal joints) include the top of the head, left ear, right ear, left eye, right eye, nose, left mouth corner, right mouth corner, head, neck, right index finger, right thumb, right palm, right wrist, right elbow, right shoulder, shoulder center, left shoulder, left elbow, left wrist, left palm, left thumb, left index finger, spine, hip center, right hip, left hip, right knee, left knee, right ankle, left ankle , right foot, left foot 33 human key points;

The selected muscle points include trapezius, pectoralis major, deltoid, trapezius, latissimus dorsi, biceps, triceps, finger extensors, serratus anterior, rectus abdominis, vastus lateralis, rectus femoris, vastus medialis, biceps femoris, gluteus maximus, gastrocnemius and soleus.

Set the adjacency matrix with a dimension of 17x25x6 according to the bipartite graph network, use the normal distribution between 0 and 1 to randomly initialize the relationship matrix, set the Huber loss function, and use the following structure to establish the graph structure:

Convolutional layer - fully connected layer - convolutional layer - fully connected layer;

The convolution layer uses the graph convolution formula:

H ^l+1 = σ(AH ^l W ^l ) (1)

where W ^l is the weight parameter matrix of layer l, and σ( ) is the Relu activation function.

A standard graph neural network training data set was established, and professional athletes were invited to attach electromyography sensing devices to the corresponding 17 muscle points to collect muscle electrical signal strength. Since the strength of the surface EMG signal at the same part of the same movement is linearly proportional to the strength of the muscle force, the surface EMG signal can be equated with the strength of the muscle force.

At the same time, the use of video equipment to collect information on the position of bone key points can record the strength of muscle electrical signals of professional athletes performing standard sports actions and the position information of bone key points obtained from image analysis. This is used as the training data for the graph neural network. Use the training data to train the adjacency matrix, optimize the parameters in the adjacency matrix, and obtain the description of the relationship between muscle force strength and bone key point movement. Through this adjacency matrix, the muscle force data can be calculated (inferred) from the position information and motion state of the bone key points.

In another embodiment of the present application, identifying the joint key points of the athlete in the video image, and obtaining the joint key point data when the athlete is moving, includes: setting the initial space coordinates of the camera, and establishing a space coordinate system based on the initial space coordinates; identifying the joint key points of the athlete in the video image, and obtaining the key point space coordinates based on the space coordinate system; establishing a first coordinate system based on any joint key points, and converting the key point space coordinates into joint key point coordinates based on the first coordinate system; according to changes in the joint key points of the athlete during movement, obtaining the joint key point data.

In the prior art, in the scheme of using a single depth camera, since the viewing distance of a single depth camera is within a 60-degree fan-shaped area of 0.8-3 meters, and the pitch angle is only 60 degrees. Therefore, the depth camera cannot capture all body parts when athletes perform large-scale sports, jumping and squatting, and due to the characteristics of the depth camera, the smart phone can only observe the position information of one side of the body. When the joints are blocked, the depth camera cannot capture the joint information, and the observation efficiency is low. Cannot provide enough information for later correction. At the same time, due to the recognition mechanism of the depth camera Kinect, there are areas of accuracy difference in the camera, and the accuracy drops seriously when it exceeds the optimal recognition area. This makes it easier for extremities to generate erroneous data, causing the motion correction process to fail.

Due to the weak correlation of the key points of the bones in the space, it is necessary to unify each key point of the bones into a coordinate system based on the tester, so as to reduce the influence of absolute displacement on the key points of the bones and increase the influence of the relative displacement of the key points of the bones and the human body on the action judgment.

Please refer to FIG. 3 , which is a schematic diagram of a method for establishing a spatial coordinate system in a sports training guidance method provided by an embodiment of the present application. In this example:

Establish a space coordinate system based on the relative positions of the three cameras;

Convert each frame pixel of the video image of the key points of the joints of the athlete into projection lines based on the three cameras respectively under the space coordinate system; calculate the mutual foot points of the projection lines; use the mean value of the mutual foot points as the key point space coordinates of the joint key points based on the space coordinate system.

Compute the real-space location to which each keypoint is mapped. Pair each key point of the three cameras separately, through the formula

The real space transformation relationship in can be obtained from the imaging model projection point q and the origin Q to obtain the projection line L of the key point in space to the camera origin.

L=λZ+R ^T t (3)

Where R ^T t is the spatial coordinates of the camera. Through L1, L2, and L3, calculate the mutual vertical points m ₁₂ , m ₂₁ , m ₁₃ , m ₃₁ , m ₂₃ , m ₃₂ , take the average value of the 6 vertical points to obtain the M point coordinates, and use the M point coordinates as joint key points based on the key point space coordinates in the space coordinate system.

In one embodiment of the present application, the key point of the neck joint is selected as the origin, the left and right shoulders are the x-axis, the vertical direction is the z-axis, the y-axis is located on the horizontal plane and perpendicular to the x-axis and z-axis, and the first coordinate system is established with the x, y, z-axis and the origin.

The 25 coordinate points relative to the camera space are transformed into the xyz coordinate system of the neck through formulas (2) (3) to generate a 5-dimensional attitude difference matrix. The matrix dimensions are (n, t, x, y, z), where n is the serial number of the key point, t is the time series number, x, y, and z are the position points of the key point in the neck coordinate system, and then obtained by comparing the front and rear frames:

This generates a new key point description matrix (n, u, v, l, x, y, z).

The above-mentioned embodiments solve the problem that the existing recognition mechanism using a single depth camera Kinect leads to a region of difference in accuracy of the camera, and the problem that the accuracy drops seriously when the best recognition region is exceeded.

It should be noted that the key point of the neck joint is selected as the origin in the above-mentioned embodiment. In the actual application process, theoretically, any key point can be used as the origin to establish the coordinate system. The purpose is to ensure the accuracy and convenience of data acquisition, which is not limited in this application.

Please refer to FIG. 4 , which is a schematic structural diagram of a stacked hourglass network algorithm in a sports training guidance method provided by an embodiment of the present application.

In this embodiment, the stacked hourglass network algorithm is used to detect each frame of the video image to identify the key points of the athlete's joints.

In this embodiment, each camera is used to capture a video image of the user's movement, and a stacked hourglass network is used to detect the key points of the whole body of the target for each frame of the video image. The stacked hourglass network is divided into two parts. The first part of the network is composed of ordinary multi-layer resnet convolution residual blocks, and finally generates the image feature map; the second part performs deconvolution operation on the feature map to obtain the target point of interest, and thus obtain the position of the key point.

In another embodiment provided by the present application, the joint key point change data of the athlete during exercise is compared with the joint key point change data of the standard movement action; according to the result of the joint key point position comparison, sports training guidance is provided.

The above-mentioned embodiment extracts the location information of the key points of the user's bones, uses the adjacency matrix of the graph neural network to infer the user's muscle force according to the position information of the key points of the bone, compares the force with the muscle force of the standard action in the database, and obtains the difference between the user's muscle force and the standard muscle force, and obtains force development suggestions and muscle node exercise suggestions; combined with the guidance method based on the position comparison of joint key points, it can provide users with comprehensive training guidance for movement and muscle force. Compared with the prior art that can only provide a simple comparison of right and wrong joint positions, the solution of the embodiment of the present application is more scientific and effective, and can provide comprehensive training guidance suggestions and improve training efficiency.

Fig. 5 is a schematic structural diagram of a sports training guidance device provided by an embodiment of the present application. As shown in FIG. 5 , the device may include a camera 200 and a central processing unit 210 .

The camera 200 is used to obtain video images of athletes;

The central processing unit may include: a joint key point recognition module 211, a muscle force processing module 212, and a sports training guidance module 213;

Joint key point identification module 211, used to identify the joint key points of the athlete in the video image, and obtain the joint key point data when the athlete is moving;

The muscle exertion processing module 212 is used to obtain the corresponding athlete's muscle point exertion data from the joint key point data according to the muscle exertion adjacency matrix model;

The sports training guidance module 213 is used to compare the athlete's muscle point force data with the standard action muscle point force data to obtain a comparison result, and provide sports training guidance according to the comparison result.

The sports training instruction device provided in this embodiment realizes the process of training instruction as follows:

The camera 200 acquires the video image of the athlete, and sends the video image information to the central processing unit 210, and the joint key point recognition module 211 in the central processing unit analyzes and processes the video image to identify the joint key point of the athlete in the video image; at the same time, according to the motion of the joint in the video image, the joint key point data of the athlete is obtained; the joint key point recognition module 211 sends the joint key point data to the muscle force processing module 212, and the muscle force processing module 212 uses the joint key point data according to the muscle force adjacency matrix model Obtain the corresponding athlete's muscle point force data and send it to the sports training guidance module 213; finally, the sports training guidance module 213 compares the athlete's muscle point force data with the standard action muscle point force data to obtain a comparison result, and provides sports training guidance according to the comparison result.

The above-mentioned embodiment obtains the video recording of the camera, detects the movement of the athlete, extracts the position information of the key points of the athlete's joints, uses the muscle force adjacency matrix model to calculate the muscle force of the athlete according to the position information of the key points of the joints, compares the muscle force data with the muscle force of the standard action, obtains the difference between the user's muscle force and the standard muscle force, and gives sports training guidance and suggestions.

Existing technical solutions are only based on a simple comparison of joint position differences, which can only provide judgments of right and wrong, but cannot provide suggestions for improving muscle strength, let alone provide comprehensive training guidance by combining movements with muscle strength. Therefore, the technical solutions provided by the embodiments of the present application effectively improve training efficiency.

At the same time, the embodiment of the present application uses an image capture method, which reduces the equipment wearing and calibration process, improves user experience, and eliminates errors between devices. Use multi-view cameras to solve the problem of blind spots caused by binocular cameras on the same side. For athletes, they can only view data and cannot receive guidance. Establish a graphic relationship network between movements and muscle strength, and provide users with professional exercise and strength guidance.

Fig. 6 shows a schematic diagram of the hardware structure of the sports training guidance device provided by the embodiment of the present application.

The motor training guidance device may include a camera 300, a processor 301 and a memory 302 storing computer program instructions.

Specifically, the processor 301 may include a central processing unit (Central Processing Unit, CPU), or a specific integrated circuit (Application Specific Integrated Circuit, ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

Memory 302 may include mass storage for data or instructions. By way of example and not limitation, memory 302 may include a Hard Disk Drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. In one example, memory 302 may include removable or non-removable (or fixed) media, or memory 302 may be a non-volatile solid-state memory. The storage 302 can be inside or outside the comprehensive gateway disaster recovery device.

In one example, the memory 302 may be a read only memory (Read Only Memory, ROM). In one example, the ROM can be mask programmed ROM, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), electrically rewritable ROM (EAROM), or flash memory or a combination of two or more of these.

Memory 302 may include read only memory (ROM), random access memory (RAM), magnetic disk storage media devices, optical storage media devices, flash memory devices, electrical, optical, or other physical/tangible memory storage devices. Thus, in general, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software comprising computer-executable instructions and which, when executed (e.g., by one or more processors), is operable to perform the operations described with reference to a method according to an aspect of the present disclosure.

The processor 301 reads and executes the computer program instructions stored in the memory 302 to implement the methods/steps S1 to S4 in the embodiment shown in FIG. 1 , and achieve the corresponding technical effects achieved by executing the methods/steps in the embodiment shown in FIG. 1 , which will not be repeated here for a brief description.

In one example, the motor training instruction device may further include a communication interface 303 and a bus 310 . Wherein, as shown in FIG. 6 , the processor 301 , the memory 302 , and the communication interface 303 are connected through a bus 310 to complete mutual communication.

The communication interface 303 is mainly used to realize the communication between various modules, devices, units and/or devices in the embodiments of the present application.

The bus 310 includes hardware, software or both, and couples the components of the online data traffic charging device to each other. By way of example and not limitation, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infiniband Interconnect, a Low Pin Count (L PC) bus, memory bus, Micro Channel Architecture (MCA) bus, Peripheral Component Interconnect (PCI) bus, PCI-Express (PCI-X) bus, Serial Advanced Technology Attachment (SATA) bus, Video Electronics Standards Association Local (VLB) bus or other suitable bus or a combination of two or more of these. Bus 310 may comprise one or more buses, where appropriate. Although the embodiments of this application describe and illustrate a particular bus, this application contemplates any suitable bus or interconnect.

It is to be understood that the application is not limited to the specific configurations and processes described above and shown in the figures. For conciseness, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art may make various changes, modifications and additions, or change the order of the steps after understanding the spirit of the present application.

The functional blocks shown in the structural block diagrams described above may be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, it may be, for example, an electronic circuit, an Application Specific Integrated Circuit (ASIC), appropriate firmware, a plug-in, a function card, and the like. When implemented in software, the elements of the present application are the programs or code segments employed to perform the required tasks. Programs or code segments can be stored in machine-readable media, or transmitted over transmission media or communication links by data signals carried in carrier waves. "Machine-readable medium" may include any medium that can store or transmit information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, Radio Frequency (RF) links, and the like. Code segments may be downloaded via a computer network such as the Internet, an Intranet, or the like.

It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, the present application is not limited to the order of the above steps, that is, the steps may be performed in the order mentioned in the embodiment, or may be different from the order in the embodiment, or several steps may be performed simultaneously.

Aspects of the present disclosure are described above with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present disclosure. It will be understood that each block of the flowchart and/or block diagrams, and combinations of blocks in the flowchart and/or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special purpose computer, or a processor of other programmable data processing devices to produce a machine, such that these instructions executed via the processor of the computer or other programmable data processing devices enable the realization of the functions/actions specified in one or more blocks of the flowcharts and/or block diagrams. Such processors may be, but are not limited to, general purpose processors, special purpose processors, application specific processors, or field programmable logic circuits. It can also be understood that each block in the block diagrams and/or flowcharts and combinations of blocks in the block diagrams and/or flowcharts can also be realized by dedicated hardware for performing specified functions or actions, or can be realized by a combination of dedicated hardware and computer instructions.

The above is only a specific implementation of the present application, and those skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the above-described system, modules and units can refer to the corresponding process in the foregoing method embodiment, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto, and any person familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should all be covered within the protection scope of this application.

Claims (9)

1. A method of athletic training coaching, comprising:

acquiring a video image of an athlete based on a camera, wherein the athlete does not need to wear an information extraction device;

identifying joint key points of the athlete in the video image, and obtaining joint key point data of the athlete during the movement;

obtaining corresponding athlete muscle point force data from the joint key point data according to a muscle force adjacency matrix model;

comparing the athlete muscle point force data with the standard action muscle point force data to obtain a comparison result, and providing exercise training guidance according to the comparison result;

the muscle development adjacency matrix model comprises:

acquiring position data of key points of a standard sport action joint and corresponding muscle point force data;

establishing a neural network training set based on the standard athletic movement joint key point position data and the corresponding muscle point force data;

establishing a bipartite graph network of the corresponding relation between the key points of the joints and the muscle points;

and optimizing a bipartite graph network of the corresponding relation between the joint key points and the muscle points according to the neural network training set to obtain the muscle stress adjacency matrix model.

2. The athletic training coaching method of claim 1, wherein identifying joint keypoints for the athlete in the video image and obtaining joint keypoint data for the athlete during the athletic activity comprises:

setting initial space coordinates of a camera, and establishing a space coordinate system based on the initial space coordinates;

identifying joint key points of athletes in the video image, and obtaining key point space coordinates based on the space coordinate system;

establishing a first coordinate system based on any joint key point, and converting the space coordinates of the key point into joint key point coordinates based on the first coordinate system;

and obtaining the joint key point data according to the joint key point change of the athlete during the exercise.

3. The method of claim 2, wherein identifying joint keypoints for the athlete in the video image and deriving spatial coordinates of the keypoints based on the spatial coordinate system comprises:

establishing a space coordinate system based on the relative positions of the three cameras;

converting each frame of pixels of the video image of the joint key points of the athlete into projection lines based on three cameras respectively under the space coordinate system;

calculating mutual foot points of the projection lines;

and taking the mean value of the mutual foot drop points as the joint key point and based on the key point space coordinates in the space coordinate system.

4. The athletic training guidance method of claim 2, wherein establishing the first coordinate system based on any joint keypoints comprises:

and selecting a neck joint key point as an origin, taking left and right shoulders as an x axis, taking a vertical direction as a z axis, taking directions which are positioned on a horizontal plane and vertical to the x axis and the z axis as y axes, and establishing a first coordinate system by using the x axis, the y axis, the z axis and the origin.

5. The athletic training coaching method of claim 1, wherein the identifying joint keypoints of the athlete in the video image comprises:

and detecting each frame of image of the video image by using a stacked hourglass network algorithm, and identifying joint key points of the athlete.

6. The athletic training coaching method of claim 1, further comprising:

the joint key point change data of the athlete during the exercise is compared with the joint key point change data of the standard exercise action;

and providing exercise training guidance according to the result of the joint key point position comparison.

7. A athletic training coaching device, the device comprising: the device comprises a camera and a central processing unit;

the camera is used for acquiring a video image of the athlete, and the athlete does not need to wear the information extraction equipment;

the central processing unit includes: the device comprises a joint key point identification module, a muscle force processing module and a motion training guidance module;

the joint key point identification module is used for identifying joint key points of the athlete in the video image and obtaining joint key point data of the athlete during the movement;

the muscle force processing module is used for obtaining corresponding athlete muscle point force data from the joint key point data according to the muscle force adjacency matrix model;

the exercise training guidance module is used for comparing the athlete muscle point force data with the standard action muscle point force data to obtain a comparison result, and providing exercise training guidance according to the comparison result;

the muscle development adjacency matrix model comprises:

acquiring position data of key points of a standard sport action joint and corresponding muscle point force data;

establishing a neural network training set based on the standard athletic movement joint key point position data and the corresponding muscle point force data;

establishing a bipartite graph network of the corresponding relation between the key points of the joints and the muscle points;

and optimizing a bipartite graph network of the corresponding relation between the joint key points and the muscle points according to the neural network training set to obtain the muscle stress adjacency matrix model.

8. An athletic training coaching device, the device comprising: a camera, a processor, and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement the athletic training coaching method according to any one of claims 1-6.

9. A computer storage medium having stored thereon computer program instructions which, when executed by a processor, implement the exercise training coaching method of any of claims 1-6.