A martial arts training monitoring method based on data analysis

Abstract

The present invention relates to the technical field of martial arts training monitoring, and specifically discloses a martial arts training monitoring method based on data analysis, comprising the following steps: S1: splitting a reference video into a number of video frames, and determining the key skeleton points of the characters in the video frames, and determining the video coordinates; S2: collecting images of students performing martial arts training at a preset position, and determining the initial body coordinates; obtaining the distance between the initial body coordinates and the origin, and obtaining the distance between the video coordinates and the origin, and calculating the correction coefficient; determining the standard body coordinates of the key skeleton points; S3: judging whether the student has positional abnormalities or angle abnormalities according to the standard body coordinates and the video coordinates. The present invention can judge whether the student's movements during martial arts training are abnormal, and improve the training effect.

Description

Technical Field

[0001] The invention relates to the technical field of martial arts training monitoring, and in particular to a martial arts training monitoring method based on data analysis. Background Art

[0002] Martial arts is a sport that has gradually formed and developed during long-term production labor, struggles with nature and wars in the era of cold weapons. It has the functions of fitness, body protection, defense against enemies and victory. It is not only a sport, but also a cultural system that includes rich philosophical, moral and aesthetic concepts. From a technical perspective, martial arts emphasizes the cultivation of both inside and outside, the unity of form and spirit, and pays attention to both the training of the external form and the cultivation of the inner spirit and breath.

[0003] At present, during martial arts training, students usually rely on the guidance of coaches to adjust martial arts movements and improve the accuracy of martial arts movements. However, such a guidance process depends largely on the personal experience and level of the coach, and there may be a problem that the coach's guidance is not accurate enough. At the same time, in actual situations, one coach usually teaches multiple students at the same time. After the coach finishes the explanation, the students usually need to practice on their own. In this process, the coach may not be able to correct the mistakes in the students' martial arts movements in time, and the students are not clear whether they have made mistakes and will continue to practice the wrong martial arts movements, which will affect the efficiency of the students' martial arts training. Summary of the invention

[0004] The purpose of the present invention is to provide a martial arts training monitoring method based on data analysis to solve the above technical problems.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A martial arts training monitoring method based on data analysis comprises the following steps:

[0007] S1: Obtain a reference video, which is a martial arts video used by students for reference, split the reference video into a number of video frames, and determine the coordinates of key skeleton points in the video frames as video coordinates, wherein the key skeleton points are the center points of preset positions on the character's body;

[0008] S2: Determine the time Tsta when the student starts martial arts training, and starting from the time Tsta, collect images of the student during martial arts training at a preset position, and determine the coordinates of key bone points in the image as initial body coordinates;

[0009] Obtain the distance Di between the initial body coordinates of the i-th key bone point and the origin, obtain the distance Di' between the video coordinates of the i-th key bone point and the origin, and calculate the correction coefficient Si = Di' / Di;

[0010] Determine the standard body coordinates BZi=(XBZi, YBZi) of the i-th key bone point, where XBZi=Xi*Si, Xi represents the abscissa of the initial body coordinates of the i-th key bone point, and YBZi=Yi*Si, Yi represents the ordinate of the initial body coordinates of the i-th key bone point;

[0011] S3: Determine whether the trainee has position abnormality or angle abnormality based on the standard body coordinates and video coordinates. The position abnormality refers to the deviation of the position of the key bone points, and the angle abnormality refers to the deviation of the angle between the line between two adjacent key bone points and the x-axis.

[0012] As a further solution of the present invention: in the step S1, the process of determining the video coordinates specifically includes:

[0013] The key skeleton points of the character in the video frame are determined by the OpenPose algorithm, a coordinate system is established with the central pixel point of the video frame as the origin, and the coordinates of the key skeleton points are determined as video coordinates.

[0014] As a further solution of the present invention: the step S2 also includes a process of preprocessing the image, and the preprocessing process specifically adjusts the resolution of the image to be consistent with the video frame.

[0015] As a further solution of the present invention: in the step S2, when collecting images of students performing martial arts training, the shooting angle is kept the same as the shooting angle corresponding to the reference video.

[0016] As a further solution of the present invention: in step S3, the process of determining whether the trainee has position abnormality specifically includes:

[0017] Obtain the standard body coordinates of the i-th key bone point at a time interval T between the time Tsta and the time interval T;

[0018] Let the time corresponding to the first video frame be Tsta, and take the video frame within the preset monitoring time [Tsta+T-T', Tsta+T+T'] as the target video frame, T' represents the preset time deviation value, and determine the video coordinates of the i-th key skeleton point in the target video frame;

[0019] Calculate position deviation Wherein, AXi' represents the abscissa of the i-th key skeleton point in the A-th target video frame, and AYi' represents the ordinate of the i-th key skeleton point in the A-th target video frame;

[0020] When the position deviation is greater than or equal to a preset position deviation threshold Wdev, it is determined to be a position abnormality.

[0021] As a further solution of the present invention: in the process of determining whether the trainee has position deviation, the position deviation set W is obtained. JH =(ΔWZ A i, ΔWZ B i, ΔWZ C i, ...), when there exists |min(W JH )-Wdev|≥Wlim, where Wlim represents the preset position deviation limit, the monitoring time is adjusted to [TKS+TT", TKS+T+T"], where T"=T'+ΔT, ΔT represents the preset time correction value.

[0022] As a further solution of the present invention: in step S3, the process of determining whether the trainee has an abnormal angle specifically includes:

[0023] Obtain the standard body coordinates of the i-th key bone point and the adjacent next key bone point i+1 at the time interval TKS, and calculate the standard angle by the formula, the formula is:

[0024]

[0025] Calculate the standard angle between the i-th key bone point and the next adjacent key bone point i+1 in the target video frame, record it as the comparison angle, and calculate the difference between the standard angle and the comparison angle, record it as the pending difference;

[0026] When the undetermined difference is greater than or equal to the preset angle deviation threshold Jdev, it is determined to be an angle abnormality.

[0027] As a further solution of the present invention: in the process of determining whether the trainee has an angle abnormality, obtain the minimum value of the pending difference, and when it is greater than or equal to the preset angle deviation limit, adjust the monitoring time to [TKS+TT", TKS+T+T"], wherein T"=T'+ΔT, ΔT represents the preset time correction value.

[0028] Beneficial effects of the present invention: In the present invention, a reference video is first obtained and processed to determine the coordinates of key skeletal points in each video frame, which provides a standard for subsequent judgment of whether there are abnormalities in the training of trainees. The coordinates of key skeletal points can provide accurate body posture data for subsequent motion analysis and correction. These coordinates can be used to quantify and compare the differences between the trainees' movements and standard movements. Key skeletal points include but are not limited to elbows, wrists, and knees. Afterwards, images of trainees performing martial arts training are collected at preset positions. The preset positions can ensure that the collected data are consistent and comparable, avoid viewing angle and distance errors caused by camera positions, and affect the accuracy of subsequent judgments. The same method is used to determine the coordinates of key skeletal points of the trainees in the image, which are used to quantify and measure whether there are abnormalities in the trainees' movements.

[0029] Next, the distance between the initial body coordinates of the key bone points and the origin is obtained, and the distance between the video coordinates of the corresponding key bone points and the origin is obtained, and the correction coefficient is calculated, and the initial body coordinates are corrected according to the correction coefficient to obtain the standard body coordinates; it is worth noting that there may be significant differences in body shape between the trainee and the person in the reference video. These differences will cause the positions of the bone points to be different even if the movements are the same. Therefore, the correction coefficient is used to eliminate the influence of the differences in body shape (such as height) between the trainee and the person in the reference video on the positions of the key bone points;

[0030] Finally, it is determined whether the trainee has position abnormality or angle abnormality according to the standard body coordinates and video coordinates; it is worth noting that position abnormality refers to the deviation of the coordinates of some skeletal points of the trainee compared with the standard (video coordinates), which is usually caused by the incorrect body posture or position of a certain part of the trainee when performing the action; and angle abnormality refers to the deviation of the angle between the line between two adjacent key skeletal points and the x-axis when the trainee performs the action, which is usually caused by improper control of the joint angle or body posture of the trainee during the action; distinguishing position abnormality and angle abnormality can improve the efficiency of the trainee's correction, thereby improving the training efficiency. The present invention can determine whether the action of the trainee during martial arts training is abnormal, and improve the training effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The present invention will be further described below in conjunction with the accompanying drawings.

[0032] Figure 1 It is a flow chart of a martial arts training monitoring method based on data analysis of the present invention. DETAILED DESCRIPTION

[0033] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] See also Figure 1 As shown, the present invention is a martial arts training monitoring method based on data analysis, comprising the following steps:

[0035] S1: Obtain a reference video, which is a martial arts video used by students for reference, split the reference video into a number of video frames, and determine the coordinates of key skeleton points in the video frames as video coordinates, wherein the key skeleton points are the center points of preset positions on the character's body;

[0036] S2: Determine the time Tsta when the student starts martial arts training, and starting from the time Tsta, collect images of the student during martial arts training at a preset position, and determine the coordinates of key bone points in the image as initial body coordinates;

[0037] Obtain the distance Di between the initial body coordinates of the i-th key bone point and the origin, obtain the distance Di' between the video coordinates of the i-th key bone point and the origin, and calculate the correction coefficient Si = Di' / Di;

[0038] Determine the standard body coordinates BZi=(XBZi, YBZi) of the i-th key bone point, where XBZi=Xi*Si, Xi represents the abscissa of the initial body coordinates of the i-th key bone point, and YBZi=Yi*Si, Yi represents the ordinate of the initial body coordinates of the i-th key bone point;

[0039] S3: Determine whether the trainee has position abnormality or angle abnormality based on the standard body coordinates and video coordinates. The position abnormality refers to the deviation of the position of the key bone points, and the angle abnormality refers to the deviation of the angle between the line between two adjacent key bone points and the x-axis.

[0040] It should be noted that, first, a reference video is obtained and processed to determine the coordinates of key bone points in each video frame. This provides a standard for subsequent judgment of whether there are abnormalities in the training of students. Determining the coordinates of key bone points can provide accurate body posture data for subsequent motion analysis and correction. These coordinates can be used to quantify and compare the differences between the student's movements and standard movements. Key bone points include but are not limited to elbows, wrists, and knees. Afterwards, images of students performing martial arts training are collected at preset positions. The preset positions can ensure that the collected data is consistent and comparable, avoid viewing angle and distance errors caused by the camera position, and affect the accuracy of subsequent judgments. The same method is used to determine the coordinates of the key bone points of the students in the image, which are used to quantify and measure whether there are abnormalities in the student's movements.

[0041] Next, the distance between the initial body coordinates of the key bone points and the origin is obtained, and the distance between the video coordinates of the corresponding key bone points and the origin is obtained, and the correction coefficient is calculated, and the initial body coordinates are corrected according to the correction coefficient to obtain the standard body coordinates; it is worth noting that there may be significant differences in body shape between the trainee and the person in the reference video. These differences will cause the positions of the bone points to be different even if the movements are the same. Therefore, the correction coefficient is used to eliminate the influence of the differences in body shape (such as height) between the trainee and the person in the reference video on the positions of the key bone points;

[0042] Finally, determine whether the trainee has position abnormality or angle abnormality based on the standard body coordinates and video coordinates; it is worth noting that position abnormality refers to the deviation of the coordinates of some of the trainee's skeletal points compared with the standard (video coordinates), which is usually caused by the incorrect body posture or position of a certain part of the trainee when performing the action; and angle abnormality refers to the deviation of the angle between the line between two adjacent key skeletal points and the x-axis when the trainee performs the action, which is usually caused by improper control of the joint angle or body posture of the trainee during the action; distinguishing between position abnormality and angle abnormality can improve the efficiency of trainees' correction, thereby improving training efficiency.

[0043] In another preferred implementation of the present invention, in step S1, the process of determining the video coordinates specifically includes:

[0044] The key skeleton points of the character in the video frame are determined by the OpenPose algorithm, a coordinate system is established with the central pixel point of the video frame as the origin, and the coordinates of the key skeleton points are determined as video coordinates.

[0045] It is worth noting that OpenPose is a deep learning model for detecting and estimating human posture. It can identify and locate the key skeletal points (also called key points) of the human body. The image to be detected is input into the model. OpenPose supports multiple input formats, such as single-frame images, videos, or image sequences captured by real-time cameras; OpenPose uses two main output branches: Key point heat maps (Part Confidence Maps): These heat maps are used to estimate the position of each key point in the image. Each key point has a corresponding heat map, in which the highlighted area indicates the possible location of the key point; Part Affinity Fields (PAF): These vector fields are used to represent the connection relationship between key points. PAF is used to describe the direction and distance information between parts of the human body (such as arms and torso); By analyzing the key point heat map, the possible position of each key point can be obtained.

[0046] In another preferred implementation of the present invention, the step S2 further includes a process of preprocessing the image, and the preprocessing process specifically adjusts the resolution of the image to be consistent with the video frame.

[0047] It is worth noting that when the resolutions of the video frame and the image are inconsistent, even if the origin of the coordinate system is set at the central pixel, the coordinates of the skeleton points will be different at different resolutions. This difference is due to the scaling caused by the change in resolution. By adjusting the resolution of the video frame to be consistent with the acquired image, the coordinate system benchmarks of the two can be ensured to be consistent, eliminating the coordinate error caused by the resolution difference. After unifying the resolution, there is no need for complex coordinate transformation, which reduces the amount of data to be processed and improves processing efficiency.

[0048] In another preferred implementation of the present invention, in the step S2, when collecting images of students performing martial arts training, the shooting angle is kept the same as the shooting angle corresponding to the reference video.

[0049] It can be understood that keeping the shooting angle the same as the shooting angle corresponding to the reference video can ensure that the skeleton point coordinates of the two are compared at the same viewing angle, prevent errors caused by different shooting angles, and improve the accuracy of judgment.

[0050] In another preferred implementation of the present invention, in step S3, the process of determining whether the trainee has abnormal position specifically includes:

[0051] Obtain the standard body coordinates of the i-th key bone point at a time interval T between the time Tsta and the time interval T;

[0052] Let the time corresponding to the first video frame be Tsta, and take the video frame within the preset monitoring time [Tsta+T-T', Tsta+T+T'] as the target video frame, T' represents the preset time deviation value, and determine the video coordinates of the i-th key skeleton point in the target video frame;

[0053] Calculate position deviation Wherein, AXi' represents the abscissa of the i-th key skeleton point in the A-th target video frame, and AYi' represents the ordinate of the i-th key skeleton point in the A-th target video frame;

[0054] When the position deviation is greater than or equal to a preset position deviation threshold Wdev, it is determined to be a position abnormality.

[0055] It should be noted that the student's movement speed may be inconsistent with the reference video. Sometimes the student may move slowly and sometimes faster. In this case, comparing at fixed time points may cause the movements to be out of sync, thus affecting the accuracy of the assessment. By comparing with each video frame during the monitoring time, even if the student's movement speed is different, it can be determined whether the student has an inaccurate position. At the same time, compared with the case of comparing with all video frames, the amount of data to be processed can be reduced, improving processing efficiency.

[0056] In another preferred implementation of the present invention, in the process of determining whether the trainee has a position deviation, a position deviation set W is obtained. JH =(ΔWZ A i, ΔWZ B i, ΔWZ C i, ...), when there exists |min(W JH )-Wdev|≥Wlim, where Wlim represents the preset position deviation limit, the monitoring time is adjusted to [TKS+TT", TKS+T+T"], where T"=T'+ΔT, ΔT represents the preset time correction value.

[0057] It should be noted that this is to prevent students from training too fast or too slow, which may cause misjudgment when the students make correct movements but the video frames corresponding to the correct movements do not fall within the monitoring time range.

[0058] In another preferred implementation of the present invention, in step S3, the process of determining whether the trainee has an abnormal angle specifically includes:

[0059] Obtain the standard body coordinates of the i-th key bone point and the adjacent next key bone point i+1 at the time interval TKS, and calculate the standard angle by the formula, the formula is:

[0060]

[0061] Calculate the standard angle between the i-th key bone point and the next adjacent key bone point i+1 in the target video frame, record it as the comparison angle, and calculate the difference between the standard angle and the comparison angle, record it as the pending difference;

[0062] When the undetermined difference is greater than or equal to the preset angle deviation threshold Jdev, it is determined to be an angle abnormality.

[0063] In another preferred implementation of the present invention, in the process of determining whether the trainee has an angle abnormality, the minimum value among the pending differences is obtained. When it is greater than or equal to the preset angle deviation limit, the monitoring time is adjusted to [TKS+TT", TKS+T+T"], wherein T"=T'+ΔT, ΔT represents the preset time correction value.

[0064] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A martial arts training monitoring method based on data analysis, characterized in that: The following steps are involved: S1: Obtain a reference video, which is a martial arts video used by students for reference, split the reference video into a number of video frames, and determine the coordinates of key skeleton points in the video frames as video coordinates, wherein the key skeleton points are the center points of preset positions on the character's body; S2: Determine the time T when students start martial arts training sta and from the time T sta Initially, an image of a student performing martial arts training is collected at a preset position, and the coordinates of key bone points in the image are determined as initial body coordinates; Get the distance D between the initial body coordinates of the i-th key bone point and the origin i , and obtain the distance D between the video coordinates of the i-th key bone point and the origin i ', and calculate the correction factor S i =D i ' / D i ; Determine the standard body coordinates BZ of the i-th key bone point i =(XBZ i , YBZ i ), where XBZ i =X i *S i , X i represents the abscissa of the initial body coordinates of the i-th key bone point, YBZ i =Y i *S i , Y i The ordinate represents the initial body coordinates of the i-th key bone point; S3: judging whether the trainee has position abnormality or angle abnormality according to the standard body coordinates and the video coordinates, wherein the position abnormality refers to the deviation of the position of the key bone points, and the angle abnormality refers to the deviation of the angle between the line between two adjacent key bone points and the x-axis; The process of determining whether a student has positional abnormality specifically includes: Get time T sta The interval between them is the standard body coordinate of the i-th key bone point at time T; Let the time corresponding to the first video frame be T sta , will be in the preset monitoring time [T sta +T-T',T sta +T+T'] as the target video frame, T' represents a preset time deviation value, and determines the video coordinates of the i-th key skeleton point in the target video frame; Calculate position deviation , where AX i ' represents the horizontal coordinate of the i-th key skeleton point in the A-th target video frame, AY i 'Shows the ordinate of the i-th key skeleton point in the A-th target video frame; When the position deviation is greater than or equal to the preset position deviation threshold W dev When , it is judged as abnormal position; In the process of judging whether the trainee has position deviation, the position deviation set W is obtained JH =( , , , …), when there exists |min(W JH )-W dev |≥W lim When W lim Indicates the preset position deviation limit, and adjusts the monitoring time to [TKS+TT", TKS+T+T"], where , Indicates the preset time correction value.

2. A martial arts training monitoring method based on data analysis according to claim 1, characterized in that: In the step S1, the process of determining the video coordinates specifically includes: The key skeleton points of the character in the video frame are determined by the OpenPose algorithm, a coordinate system is established with the central pixel point of the video frame as the origin, and the coordinates of the key skeleton points are determined as video coordinates.

3. A martial arts training monitoring method based on data analysis according to claim 1, characterized in that: The step S2 also includes a process of preprocessing the image, and the preprocessing process specifically adjusts the resolution of the image to be consistent with the video frame.

4. The method for monitoring martial arts training based on data analysis according to claim 1, characterized in that: In the step S2, during the process of collecting images of students performing martial arts training, the shooting angle is kept the same as the shooting angle corresponding to the reference video.

5. The method for monitoring martial arts training based on data analysis according to claim 1, characterized in that: In step S3, the process of determining whether the trainee has an angle abnormality specifically includes: Obtain the standard body coordinates of the i-th key bone point and the adjacent next key bone point i+1 at the time interval TKS, and calculate the standard angle by the formula, the formula is: ; Calculate the standard angle between the i-th key bone point and the next adjacent key bone point i+1 in the target video frame, record it as the comparison angle, and calculate the difference between the standard angle and the comparison angle, record it as the pending difference; When the undetermined difference is greater than or equal to the preset angle deviation threshold J dev , it is judged as an angle abnormality.

6. A martial arts training monitoring method based on data analysis according to claim 5, characterized in that: In the process of judging whether the trainee has an angle abnormality, the minimum value of the pending difference is obtained. When it is greater than or equal to the preset angle deviation limit, the monitoring time is adjusted to [TKS+TT", TKS+T+T"], where , Indicates the preset time correction value.

Images