WO2023151200A1

WO2023151200A1 - Action-recognition-based human body activity level measurement method, system and apparatus, and medium

Info

Publication number: WO2023151200A1
Application number: PCT/CN2022/094627
Authority: WO
Inventors: 曾晓嘉; 刘易; 薛立君
Original assignee: 成都拟合未来科技有限公司
Priority date: 2022-02-10
Filing date: 2022-05-24
Publication date: 2023-08-17
Also published as: CN116631045A

Abstract

The present disclosure relates to an action-recognition-based human body activity level measurement method, system and apparatus, and a medium, and relates to the field of fitness. The method comprises: collecting images of a user in an area within a preset time of t seconds, so as to obtain consecutive action image frames of the user; according to the consecutive action image frames, acquiring the average speed of the user within t seconds; and according to the average speed of the user, acquiring an activity level of the user within t seconds, and displaying the activity level. In the present disclosure, the average speed of a user is acquired by means of consecutive action image frames of the user, and an activity level of the user when an action occurs is acquired according to the average speed, such that the acquired activity level is more precise; and for a small-amplitude action, the average speed of the user can also be more accurately acquired by means of the consecutive action image frames, thereby reducing the error.

Description

Human activity detection method, system, device, and medium based on motion recognition

technical field

The present disclosure relates to the field of fitness, in particular to a method, system, device, and medium for detecting human activity based on motion recognition.

Background technique

With the improvement of living standards, people pay more and more attention to the pursuit of healthy life. With the development and progress of science and technology, the existing fitness equipment continues to develop in the direction of familyization and intelligence. Under this premise, intelligent mirror fitness equipment gradually begins to appear in people's lives. The intelligent mirror fitness device uses various equipment built into the body and is displayed and mirrored by the front mirror. Before this case, the applicant has carried out a lot of research on the fitness mirror related solutions and filed a patent application, including the effective recognition of user actions , feedback on user actions, etc. With the continuous refinement of the research, the applicant found that in the process of the development of the fitness device towards the home, the user's enthusiasm for exercise is very important. When using the home fitness device for fitness, It is necessary to mobilize the user's fitness enthusiasm so that the user can achieve the exercise goal. Based on this, the applicant found that by feeding back the amount of exercise to the user, that is, the activity of the exercise can better achieve the effect of motivating the user to achieve the exercise goal. However, most of the existing activity detection methods use wearable devices to detect, but This detection method can only roughly detect the activity of the user, and is not suitable for real-time detection during the fitness process, so as to achieve the effect of motivating the user.

Contents of the invention

The disclosure provides a human activity detection method, system, device, and medium based on motion recognition. The purpose of the disclosure is to obtain the user's average speed through continuous frame motion images of the user, and obtain the user's activity when an action occurs based on the average speed.

To achieve the above purpose, the present disclosure provides a method for detecting human activity based on motion recognition, including:

Within the preset time T seconds, collect images of users in the area to obtain continuous frames of motion images of users;

Obtain the average speed of the user within T seconds according to the continuous frame action images;

Obtain user activity within T seconds according to the user's average speed, and display the activity.

The existing method for obtaining activity degree collects the acceleration data of the human body through the terminal, identifies the action type based on the collected acceleration data, and calculates the human activity index based on the identified action type and the metabolic equivalent corresponding to the action type obtained by looking up the table. This disclosure obtains the user's average speed through each frame of action images, and obtains the activity degree based on the user's average speed. For small movements, the existing methods cannot accurately reflect the user's activity degree, but the present disclosure obtains the user's activity degree through continuous frames. Action images can more accurately reflect user activity.

There are many ways to obtain the user's average speed. It can be through the coordinate change of one or more bone points in the action image, and a position that can best reflect the user's action range can be obtained according to the user's action. This position can be an action A certain point in the image can also be a line segment or a surrounding graphic, as long as it can accurately reflect the state of the user's movement, and finally get the user's average speed. Reflect user activity.

On this basis, in order for the applicant to find out that the user's average speed can be better reflected by determining the user's average speed through the user's skeletal points in the action image, especially for local actions such as hand movements or leg movements, etc. In other words, this method can directly obtain the speed that best reflects the user's actions, including:

Obtain several bone points of the user in each action image and the information corresponding to each bone point;

Obtain the average speed of each bone point within T seconds according to the information corresponding to each bone point;

Get the user's average speed based on the average speed of each bone point.

Among them, the information corresponding to each bone point includes the coordinates corresponding to each bone point, and the average speed of the jth bone point within T seconds:

Where n is the number of action image frames included in T seconds, (xi _, y _i ) is the coordinate of the skeleton point in the i-th action image, and j and i are both positive integers. In this application, there are j bone points in total, and j can be 16, 12 or 14, which is determined according to actual usage conditions. After determining the average speed of each skeletal point within T seconds according to the above formula, since for an action, the proportion of each skeletal point affecting the user's average action is different in each action, the present disclosure obtains each skeletal point After the average speed of each bone point, the final average speed of the user is also obtained according to the preset weight of each bone point. Specifically, each bone point is preset with a weight w _j , j is a positive integer, and the average speed of the user is:

Among them, v _j is the average velocity of the j-th bone point within T seconds, w _j is the preset weight of the j-th bone point, and m is the number of bone points of the user in the action image. In this application, the weight of each skeletal point can be determined according to the type of each action, and can also be obtained through induction. In this disclosure, the specific size of the weight is based on the actual situation or the type of action As long as it can reflect the proportion of each bone point in the user's action image and clearly and accurately reflect the user's action.

In this disclosure, preferably, there are 16 skeletal points in the user's action image, including head, neck, throat, crotch, left hand, right hand, left elbow, right elbow, left foot, right foot, left shoulder , right shoulder, left knee, right knee, left hip, and right hip, where the weight of head, neck, throat, and hip is 1, the weight of left hand, right hand, left elbow, and right elbow is 1.5, and the weight of left foot, right Weight 2.5 for foot, left shoulder, right shoulder, weight 4 for left knee, right knee, weight 4.5 for left hip, right hip.

After obtaining the average speed of the user based on the average speed of each skeletal point through the above formula, the applicant also found that when collecting user images in the area, the user stands closer and farther away for the same action, because The image has a relationship between nearness, distance, and distance. When the station is near, the pixel distance passed by the skeleton point is greater, and the calculated average speed V of the user will be greater. Therefore, in order to keep the calculated speed invariant to the distance, the applicant obtains the relative average speed of the user according to the average speed of the user and the preset reference distance of the human body

According to the relative average speed of users

Get user activity within T seconds. Specifically, the preset human body reference distance is to select the human body reference distance S, and the human body reference distance S is obtained through two bone points in the upper body. The selection of the two bone points can be the top bone point and the pelvic bone point, or Throat bone points and hip bone points, etc. The reference point used to determine the reference distance of the human body is stable through the distance length between the two reference points, so that when the user stands far or close, the relative average speed of the user can be accurately obtained through the scale

Preferably, in the present disclosure, the throat bone point and the crotch bone point are selected as two reference points, and the human body reference distance S is obtained by calculating the Euclidean distance between the throat bone point and the crotch bone point, and then through the user's The ratio of the average speed V to the reference distance S of the human body is used to obtain the relative average speed of the user

Get the user's relative average speed through several bone points

After that, you need to add

It is converted into a more intuitive data, which makes it easier for users to intuitively see their current activity during exercise. Usually, the score of 0-100 can most intuitively reflect the user's activity for users, so in In this disclosure, the user's relative average speed

Converted to a score of 0-100, used to indicate the user's current activity. Specifically include:

Preset the mapping function of user average speed and activity score;

Get the liveness score according to the obtained average speed of the user and the mapping function.

In this disclosure, the mapping function can be a linear function or a nonlinear function. When the mapping function is preset, when the user's relative average speed

When it is 0, the activity score is 0, when the user's relative average speed

When greater than or equal to the preset relative average speed maximum, the liveness score is 100.

In this disclosure, there are three preset mapping functions, which are the first nonlinear function, the second nonlinear function and the linear function. When selecting the mapping function, choose according to the actual situation. If the movement of bone points increases evenly and linearly, choose linear mapping; if you want the user's score to rise rapidly as the movement speeds up, choose the first nonlinear function; if you want it difficult for the user to achieve a high-score exercise, choose the second non-linear function.

Among them, the first nonlinear function is:

The second nonlinear function is:

The first nonlinear function is:

Where Vmax is a preset relative average speed maximum value.

Corresponding to the method in the present disclosure, the present disclosure also provides a human activity detection system based on motion recognition, including:

The collection module is used to collect user images in the area within the preset time T seconds to obtain continuous frames of motion images of the user;

The first calculation module is used to obtain the average speed of the user in T seconds according to the continuous frame action images;

The second calculation module is used to obtain user activity within T seconds according to the average speed of the user;

The display module is used to display user activity.

The bone point module is used to obtain several bone points of the user in each action image and the corresponding coordinates of each bone point.

The motion recognition-based human activity detection system of the present disclosure can perform operations corresponding to the motion recognition-based human body activity detection method.

Corresponding to the method in the present disclosure, the present disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and operable on the processor, and the processor executes the The computer program realizes the above-mentioned human activity detection method based on motion recognition.

Corresponding to the method in the present disclosure, the present disclosure also provides a storage medium, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned human activity detection method based on motion recognition is realized operation.

One or more technical solutions provided by the present disclosure have at least the following technical effects or advantages:

This disclosure obtains the average speed of the user through the user's continuous frame action images, and obtains the user's activity when the action occurs according to the average speed, and the acquired activity is more accurate. Accurately obtain the average speed of users and reduce errors.

At the same time, the disclosure converts the activeness into scores for display, and shows the activeness of the current exercise to the user during the user's exercise, which can better motivate the user to exercise and improve enthusiasm.

Description of drawings

The drawings described here are used to provide further understanding of the embodiments of the present disclosure, constitute a part of the application, and do not limit the embodiments of the present disclosure. In the attached picture:

FIG. 1 is a schematic flow chart of a human activity detection method based on motion recognition;

FIG. 2 is a schematic diagram of the composition of a human activity detection system based on motion recognition.

Detailed ways

In order to understand the above-mentioned purpose, features and advantages of the present disclosure more clearly, the present disclosure will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, under the condition of not conflicting with each other, the embodiments of the present disclosure and the features in the embodiments can be combined with each other.

In the following description, many specific details are set forth in order to fully understand the present disclosure. However, the present disclosure can also be implemented in other ways different from the scope of this description. Therefore, the protection scope of the present disclosure is not limited by the following disclosure. limitations of specific examples.

Those skilled in the art should understand that, in the disclosure of the present disclosure, the terms "vertical", "transverse", "upper", "lower", "front", "rear", "left", "right", " The orientation or positional relationship indicated by "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, which are only for the convenience of describing the present disclosure and The above terms are not to be construed as limiting the present disclosure because the description is simplified rather than indicating or implying that the device or element referred to must have a particular orientation, be constructed, and operate in a particular orientation.

It can be understood that the term "a" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element The quantity can be multiple, and the term "a" cannot be understood as a limitation on the quantity.

Embodiment one

Please refer to FIG. 1. FIG. 1 is a schematic flowchart of a method for detecting human activity based on motion recognition. The present disclosure provides a method for detecting human activity based on motion recognition. The method includes:

In this embodiment, there are many ways to obtain the user's average speed within T seconds through continuous frames of motion images, which can be to determine one key point or multiple key points or between key points in the motion image according to the user's action type. A coordinate system is established in the action image, and within T seconds, the average speed of the user within T seconds is obtained through the position transformation of the key point in the continuous frames of the action image.

In this embodiment, there is a preset corresponding relationship between the average speed of the user and the degree of activity in this embodiment. After obtaining the average speed of the user, the degree of activity can be directly obtained according to the preset corresponding relationship, and displayed, thereby motivating the user .

Embodiment two

On the basis of Embodiment 1, this embodiment selects multiple skeleton points as key points to obtain the average speed of the user, specifically including:

The information corresponding to each bone point includes the coordinates corresponding to each bone point, and the average speed of the jth bone point within T seconds:

Where n is the number of action image frames included in T seconds, (xi _, y _i ) is the coordinate of the skeleton point in the i-th action image, and j and i are both positive integers.

Obtain the user's average speed according to the average speed of each bone point;

Each skeletal point is preset with a weight w _j , j is a positive integer, and the user's average speed:

Among them, the average speed of the j-th bone point within T seconds, w _j is the preset weight of the j-th bone point, and m is the number of bone points of the user in the action image.

In this embodiment, several skeleton points can be 12 skeleton points, can also be 14 skeleton points, can also be 16 skeleton points, in this embodiment, the number and position of the skeleton points are not limited, as long as they can Representing the movement of the hands, legs and head respectively.

The method for detecting human activity based on motion recognition in the present disclosure will be introduced in conjunction with specific examples below:

Operation 1 collects user images in the area within 1 second of the preset time, and obtains 25 consecutive frames of user's action images;

Operation 2 obtains several bone points of the user in each action image and information corresponding to each bone point;

Operation 2.1 Set the lower left corner of the action image as the origin, the horizontal direction as the x-axis, the vertical direction as the y-axis, and the unit length as 1 pixel to establish a rectangular coordinate system, and use AI to locate the 16 bone points and 16 bone points of the user in the action image. Coordinates of bone points, including head, neck, throat, crotch, left hand, right hand, left elbow, right elbow, left foot, right foot, left shoulder, right shoulder, left knee, right knee, left hip, and right hip.

Operation 3 Obtain the average speed of each bone point within 1 second according to the information corresponding to each bone point;

The average speed of the jth bone point within 1 second:

Where n is the number of action image frames included in 1 second, in this embodiment n is 25, ( _xi , y _i ) is the coordinates of the skeleton point in the i-th action image, j and i are both positive integers . Head, neck, throat, crotch, left hand, right hand, left elbow, right elbow, left foot, right foot, left shoulder, right shoulder, left knee, right knee, left buttock and right buttock in order, that is, the head is the serial number 1. The right hip is number 16, v ₁ is the average speed of the head bone points, v ₁₆ is the average speed of the right hip bone points, and the average of 25 frames of motion images in 1 second for 16 bone points is calculated by formula 1 Velocity v _j .

Operation 4 obtains the user's average speed according to the average speed of each bone point;

Wherein, the average speed of the jth bone point within 1 second, _wj is the preset weight of the jth bone point, m is the number of user's bone points in the action image, and m is 16 in this embodiment.

In this embodiment, the weights of the head, neck, throat, and crotch are 1, the weights of the left hand, right hand, left elbow, and right elbow are 1.5, the weights of the left foot, right foot, left shoulder, and right shoulder are 2.5, and the weights of the left and right hands are 1.5. Knee, right knee weight 4, left hip and right hip weight 4.5.

Operation 5 Obtain the user's relative average speed according to the user's average speed v and the preset human reference distance s

In this embodiment, the reference distance s of the human body is determined by the bone point of the throat and the bone point of the crotch, the coordinates of the bone point of the throat are (x1, y1), and the coordinates of the body are (x2, y2), then the reference pixel distance s:

S=sqrt((x1-x2)*(x1-x2)+(y1-y2)*(y1-y2)), that is, the common two-point Euclidean distance.

Operation 6 according to the user's relative average speed

Get activity score;

Operation 6.1 Preset the mapping function of user average speed and activity score;

The first nonlinear function is:

The second nonlinear function is:

The first nonlinear function is:

Where Vmax is the preset relative average speed maximum value;

Operation 6.2 Calculate the relative average speed of users

As x is brought into the mapping function in operation 6.1, when the user's relative average speed

In this embodiment, the first non-linear function is selected as the mapping function to obtain activity, and after the activity is obtained, the activity score is displayed.

In this embodiment, the device is a smart fitness mirror. The body of the smart fitness mirror has built-in various equipment, and a mirror surface for display and mirror display is provided on the front. When the user is exercising, the mirror surface of the smart fitness mirror can display the activity level.

In this embodiment, the user continuously displays the activity level on the mirror surface of the smart fitness mirror during the exercise. In this embodiment, T=1 second, including 25 frames of action images per second, assuming that the continuous frame numbers of a certain period of time are 1, 2, 3, 4, 5, 6, ... 100, a total of about 4 seconds In the time, calculate the average speed of each bone point in 1-25 frames, 2-26 frames, 3-27 frames, ... 76-100 frames. That is, at each moment, the average speed of each bone point in the first 25 frames at that moment can be calculated. In turn, the activity of the user can be displayed at every moment to achieve a better incentive effect.

Embodiment three

Please refer to FIG. 2. FIG. 2 is a schematic diagram of the composition of a human activity detection system based on motion recognition. Embodiment 3 of the present disclosure provides a human body activity detection system based on motion recognition. On the basis of Embodiment 1 or 2, the The system includes:

The collection module is used to collect user images in the area within the preset time T seconds to obtain continuous frame action images of the user; the collection module includes a camera, and in this embodiment, the user image is obtained through the camera;

The display module is used to display user activity.

Wherein, the first calculation module is used to obtain the average speed of the user within T seconds according to several skeleton points in the continuous frames of action images and the coordinates corresponding to each skeleton point.

The system for detecting human activity based on motion recognition in this embodiment may perform operations corresponding to the method for detecting human activity based on motion recognition in Embodiment 1 and/or Embodiment 2.

Embodiment four

Embodiment 4 of the present disclosure provides an electronic device, including a memory, a processor, and a computer program stored in the memory and operable on the processor. When the processor executes the computer program, the Human activity detection method based on action recognition.

Wherein, the processor may be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits, off-the-shelf programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components etc. A general-purpose processor may be a microprocessor, or the processor may be any conventional processor, or the like.

The memory can be used to store the computer programs and/or modules, and the processor can implement various functions of the motion recognition-based human activity detection device in the present disclosure by running or executing the data stored in the memory. The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, at least one application required by a function (such as a sound playback function, an image playback function, etc.) and the like. In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as hard disk, internal memory, plug-in hard disk, smart memory card, secure digital card, flash memory card, at least one magnetic disk storage device, flash memory device, or other volatile solid-state memory devices.

Embodiment five

Embodiment 5 of the present disclosure provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the operation of the method for detecting human activity based on motion recognition is realized. .

The computer storage medium in the embodiments of the present disclosure may use any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (non-exhaustive list) of computer-readable storage media include: electrical connections with one or more conductors, portable computer disks, hard disks, random access memory (RAM), read-only memory (ReadOnlyMemory, ROM ), erasable programmable read-only memory ((ErasableProgrammableReadOnlyMemory, EPROM) or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

While preferred embodiments of the present disclosure have been described, additional changes and modifications can be made to these embodiments by those skilled in the art once the basic inventive concept is appreciated. Therefore, it is intended that the appended claims be construed to cover the preferred embodiment and all changes and modifications which fall within the scope of the present disclosure.

It is obvious that those skilled in the art can make various changes and modifications to the present disclosure without departing from the spirit and scope of the present disclosure. Thus, if these modifications and variations of the present disclosure fall within the scope of the claims of the present disclosure and equivalent technologies thereof, the present disclosure also intends to include these modifications and variations.

Claims

Human activity detection method based on action recognition, including:

Within the preset time T seconds, collect images of users in the area to obtain continuous frames of motion images of users;

Obtain the average speed of the user within T seconds according to the continuous frame action images;

Obtain user activity within T seconds according to the user's average speed, and display the activity.
The human body activity detection method based on motion recognition according to claim 1, wherein the average speed of the user within T seconds is obtained according to continuous frame motion images, specifically comprising:

Obtain several bone points of the user in each action image and the information corresponding to each bone point;

Obtain the average speed of each bone point within T seconds according to the information corresponding to each bone point;

Get the user's average speed based on the average speed of each bone point.
The human body activity detection method based on action recognition according to claim 2, wherein the information corresponding to each skeletal point includes the coordinates corresponding to each skeletal point,

The average speed of the jth bone point within T seconds:

n is the number of action image frames included in T seconds, ( xi , y i ) is the coordinate of the skeleton point in the i-th action image, j and i are both positive integers.
The human activity detection method based on action recognition according to claim 2 or 3, wherein each skeletal point is preset with a weight w j , j is a positive integer, and the average speed of the user is:

v j is the average velocity of the j-th bone point within T seconds, w j is the preset weight of the j-th bone point, and m is the number of bone points of the user in the action image.
The human body activity detection method based on motion recognition according to any one of claims 2-4, wherein the user's relative average speed is obtained according to the user's average speed and the preset human body reference distance, and T seconds is obtained according to the user's relative average speed internal user activity.
The human body activity detection method based on action recognition according to claim 1, wherein the user activity within T seconds is obtained according to the user's average speed, specifically comprising:

Preset the mapping function of user average speed and activity score;

Get the liveness score according to the obtained average speed of the user and the mapping function.
Human activity detection system based on motion recognition, including:

The collection module is used to collect user images in the area within the preset time T seconds to obtain continuous frames of motion images of the user;

The first calculation module is used to obtain the average speed of the user in T seconds according to the continuous frame action images;

The second calculation module is used to obtain user activity within T seconds according to the average speed of the user;

The display module is used to display user activity.
The human activity detection system based on action recognition according to claim 7, further comprising:

The bone point module is used to obtain several bone points of the user in each action image and the corresponding coordinates of each bone point.
The human activity detection system based on action recognition according to claim 7, wherein the first calculation module specifically executes:

Obtain several bone points of the user in each action image and the information corresponding to each bone point;

Obtain the average speed of each bone point within T seconds according to the information corresponding to each bone point;

Get the user's average speed based on the average speed of each bone point.
The human activity detection system based on action recognition according to claim 9, wherein the information corresponding to each skeletal point includes the coordinates corresponding to each skeletal point,

The average speed of the jth bone point within T seconds:

n is the number of action image frames included in T seconds, ( xi , y i ) is the coordinate of the skeleton point in the i-th action image, j and i are both positive integers.
The human body activity detection system based on motion recognition according to claim 9 or 10, wherein,

Each skeletal point is preset with a weight w j , j is a positive integer, and the user's average speed:

v j is the average velocity of the j-th bone point within T seconds, w j is the preset weight of the j-th bone point, and m is the number of bone points of the user in the action image.
The human activity detection system based on motion recognition according to any one of claims 9-11, wherein,

The second calculation module obtains the user's relative average speed according to the user's average speed and the preset human body reference distance, and obtains the user's activity within T seconds according to the user's relative average speed.
The human body activity detection system based on action recognition according to claim 7, wherein the second calculation module acquires user activity within T seconds according to the user's average speed, specifically comprising:

Preset the mapping function of user average speed and activity score;

Get the liveness score according to the obtained average speed of the user and the mapping function.
A storage medium, the computer-readable storage medium stores a computer program, wherein, when the computer program is executed by a processor, the following operations of the human body activity detection method based on motion recognition are realized:

Within the preset time T seconds, collect images of users in the area to obtain continuous frames of motion images of users;

Obtain the average speed of the user within T seconds according to the continuous frame action images;

Obtain user activity within T seconds according to the user's average speed, and display the activity.
The storage medium according to claim 14, wherein obtaining the average speed of the user within T seconds according to the continuous frames of action images specifically includes:

Obtain several bone points of the user in each action image and the information corresponding to each bone point;

Obtain the average speed of each bone point within T seconds according to the information corresponding to each bone point;

Get the user's average speed based on the average speed of each bone point.
The storage medium according to claim 15, wherein the information corresponding to each skeletal point includes the coordinates corresponding to each skeletal point,

The average speed of the jth bone point within T seconds:

n is the number of action image frames included in T seconds, ( xi , y i ) is the coordinate of the skeleton point in the i-th action image, j and i are both positive integers.
The storage medium according to claim 15 or 16, wherein each skeletal point is preset with a weight w j , j is a positive integer, and the average speed of the user is:

v j is the average velocity of the j-th bone point within T seconds, w j is the preset weight of the j-th bone point, and m is the number of bone points of the user in the action image.
The storage medium according to any one of claims 15-17, wherein the user's relative average speed is obtained according to the user's average speed and the preset human body reference distance, and the user's activity within T seconds is obtained according to the user's relative average speed.
The storage medium according to claim 14, wherein obtaining user activity within T seconds according to the average speed of the user specifically includes:

Preset the mapping function of user average speed and activity score;

Get the liveness score according to the obtained average speed of the user and the mapping function.