A martial arts action scoring method, system and storage medium based on artificial intelligence

By shooting multi-angle human body images and comparing features, three-dimensional action videos are generated for trajectory scoring, which solves the subjectivity and accuracy of martial arts action scoring, and achieves more accurate and objective scoring.

CN118470750BActive Publication Date: 2024-11-08GUANGDONG UNIV OF SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410698481.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-31
Publication Date
2024-11-08
Estimated Expiration
2044-05-31

AI Technical Summary

Technical Problem

The existing martial arts movement scoring methods have problems such as high subjectivity and low scoring accuracy, especially due to the influence of different human characteristics, it is impossible to accurately compare the movements of the target object with the standard movements.

Method used

The camera unit takes human images from multiple angles of the target object, extracts the target human characteristics, and compares them with the stored standard human characteristics, obtains the closest standard action video, generates three-dimensional action video, performs trajectory comparison, and combines the action completion time score. Get the final rating.

Benefits of technology

It improves the accuracy and objectivity of martial arts movement scores, takes into account the standards and fluency of the movements, and is suitable for martial arts movement scores of different heights and sizes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118470750B_ABST
    Figure CN118470750B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of data processing technology, and more specifically to a martial arts action scoring method, system and storage medium based on artificial intelligence. The method comprises: step S1: extracting the target human body features based on a human body image; step S2: comparing the target human body features with a plurality of standard human body features stored in a storage unit, obtaining the first human body features and the second human body features closest to the target human body features, and obtaining the first standard action video and the second standard action video; step S3: comparing the target action video with the first standard action video and the second standard action video, and obtaining the first score based on the first comparison result; step S4: obtaining the second score based on the completion time of each sub-scoring action; step S5: obtaining the final score based on the first score and the second score. The present invention solves the problem of intelligent scoring of martial arts action scoring and improves the accuracy of martial arts action scoring.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and more specifically to a martial arts action scoring method, system and storage medium based on artificial intelligence. Background Art

[0002] Traditional martial arts are the treasure of the Chinese nation. Many people learn traditional martial arts to promote the Chinese spirit and cultivate their character. The scoring of traditional martial arts movements is done by martial arts referees based on the technical level of the athletes on the spot and the movement specifications of each competition item. This is quite subjective and cannot obtain the entire martial arts movement from multiple angles, which makes the scoring one-sided. Existing technologies, such as Chinese patent CN107341612A, involve a movement evaluation method for rehabilitation training based on Tai Chi. Due to the provision of an expert model and a trainer model, a standard Tai Chi is created based on the Tai Chi movements of Tai Chi experts. Tai Chi animation model library, trainees follow the standard Tai Chi animation model library to learn Tai Chi to perform rehabilitation training, and the trainees' body movements are mapped to the trainee model in real time; and training scores and evaluation results are given according to the matching degree between the trainee model's movements and the movements of the expert model preset in the standard Tai Chi animation model library, which can well combine traditional Tai Chi with virtual reality rehabilitation training, which is beneficial to improving the rehabilitation training effect of trainees. For example: Japanese patent JP6082101B2, the present invention provides a body movement scoring device that can score body movements that are not easily affected by the physique differences of the scored person. A body position information acquisition device (11) is used to acquire body position information of the scoring object person in time series, identify the position of each part in the acquired body position information in time series, and use other parts as a reference. A body part recognition device (12) is used to relatively identify the position of the body part, and the position of the body part recognized by the body part recognition device is compared with a pre-prepared comparison condition in time sequence. A body movement scoring device (1) comprises a body movement scoring device (13) and a scoring result display device (14) for displaying the scoring result. Both of the above patents solve the problem of action scoring. Since objects with different body characteristics have different action trajectories when performing the same action, the above two patent documents do not consider comparing the target action of the target object with a reference action that is similar to its body characteristics. Therefore, the accuracy of the obtained action score is not high enough. Summary of the invention

[0003] In order to better solve the above problems, the present invention provides a martial arts action scoring method based on artificial intelligence, the method comprising the following steps:

[0004] Step S1: photographing human body images of a target object at multiple angles within a first preset time by a camera unit, and extracting the target human body features based on the human body images;

[0005] Step S2: Compare the target human body feature with a plurality of standard human body features stored in a storage unit, obtain a first human body feature and a second human body feature that are closest to the target human body feature, and obtain a first standard action video corresponding to the first human body feature and a second standard action video corresponding to the second human body feature based on the first human body feature and the second human body feature;

[0006] Step S3: using the camera unit to capture real-time action videos of the target object at N angles in real time within a second preset time, and generating a three-dimensional target action video based on the real-time action video, placing the target action video, the first standard action video, and the second standard action video in the same three-dimensional space, and comparing the target action video with the first standard action video and the second standard action video, and obtaining a trajectory comparison result, and obtaining a first score based on the trajectory comparison result;

[0007] Step S4: according to the completion time of each sub-scoring action of the preset martial arts action in the target action video and the standard completion time of the sub-scoring action, based on the time difference between the completion time and the standard completion time, obtain the sub-action score of each sub-scoring action, and obtain the second score of the preset martial arts action according to each sub-action score;

[0008] Step S5: Obtain a final score based on the first score and the second score.

[0009] As a preferred technical solution of the present invention, the target human body features include the body shape of the target object, and the body shape includes height, fatness and thinness, and the proportion of each body part to the entire body. The first standard video and the second standard video are three-dimensional videos.

[0010] As a preferred technical solution of the present invention, step S2 includes the following steps:

[0011] Step S21: reading a plurality of the standard human features from a storage unit, and comparing the target human feature with the height, fatness, and proportion of each body part in the whole body in each of the standard human features, and obtaining a first comparison result, a second comparison result, and a third comparison result respectively, and weighting the first comparison result, the second comparison result, and the third comparison result to obtain a final comparison result, wherein the first comparison result has the highest weight, the third comparison result has the second highest weight, and the second comparison result has the lowest weight, and obtaining the final comparison results corresponding to all the standard human features, and taking the standard human features corresponding to the two smallest final comparison results as the first human feature and the second human feature;

[0012] Step S22: reading the first standard action video corresponding to the first human feature and the second standard action video corresponding to the second human feature from the storage unit;

[0013] Among them, the height in any one of the first human feature and the second human feature is higher than the height in the target human feature, and the height in the other one is lower than the height in the target human feature.

[0014] As a preferred technical solution of the present invention, step S3 includes the following steps:

[0015] Step S31: using the multiple cameras of the camera unit to capture real-time action videos of the target object at N angles in real time within the second preset time, where N is a positive integer greater than or equal to 3;

[0016] Step S32: generating the target action video according to the real-time action videos at N angles, and placing the target action video, the first standard action video and the second standard action video in the same three-dimensional space, wherein the center of gravity of the target object in the target action video is in a straight line with the center of gravity of the first object corresponding to the first standard action video and the second object in the second standard action video, and the feet of the target object are in the same horizontal line with the feet of the first object and the second object;

[0017] Step S33: Based on the first standard action video and the second standard action video, respectively obtain the first action trajectory and the second action trajectory of the preset martial arts action, obtain the target action trajectory of the preset martial arts action through the target action video, compare the target action trajectory with the first action trajectory and the second action trajectory, and obtain the trajectory comparison result. When the trajectory comparison result is that the target action trajectory falls entirely between the first action trajectory and the second action trajectory, the first score is full marks.

[0018] As a preferred technical solution of the present invention, when the first comparison result is that the target motion trajectory is not all in the area between the first motion trajectory and the second motion trajectory, the target motion trajectory is divided into a plurality of target sub-scoring motion trajectories, and the first motion trajectory and the second motion trajectory are respectively divided into a plurality of first sub-scoring motion trajectories and a plurality of second sub-scoring motion trajectories, wherein each of the target sub-scoring motion trajectories corresponds to each of the first sub-scoring motion trajectory and the second sub-scoring motion trajectory, respectively, and when the target sub-scoring motion trajectory falls between the first sub-scoring motion trajectory and the second sub-scoring motion trajectory, the target sub-scoring motion trajectory corresponds to The sub-action score is full marks; when the target sub-scoring action trajectory falls on both sides of the first sub-scoring action trajectory and the second sub-scoring action trajectory, the deviation value between the target sub-scoring action trajectory and the first sub-scoring action trajectory or the second sub-scoring action trajectory that is closer to the target sub-scoring action trajectory is calculated, and when the deviation value is less than or equal to a first threshold, the sub-action score corresponding to the target sub-scoring action trajectory is full marks; when the deviation value is greater than the first threshold, the sub-action score is obtained according to the deviation value, wherein the larger the deviation value, the smaller the sub-action score, and vice versa, the larger the sub-action score, and the first score is obtained based on each sub-action score.

[0019] As a preferred technical solution of the present invention, step S4 includes the following steps:

[0020] Step S41: dividing the preset martial arts action into a plurality of sub-scoring actions, and respectively calculating the completion time of each of the sub-scoring actions in the target action video;

[0021] Step S42: Compare the completion time with the standard completion time of the sub-scoring action to obtain the time difference, and obtain the sub-action score of the sub-scoring action based on the time difference and the scoring standard, and obtain the second score of the preset martial arts action based on each sub-action score.

[0022] As a preferred technical solution of the present invention, the step S5 includes: calculating the final score according to the weight of the first score and the weight of the second score, wherein the first score weight is greater than the second score weight.

[0023] As a preferred technical solution of the present invention, the step S3 also includes, when the speed of the preset martial arts movement in the target action video is inconsistent with the speed of the preset martial arts movement in the first standard action video and the second standard action video, adjusting the action speed of the first standard action video and the second standard action video so that the speed of the preset martial arts movement in the first standard action video and the second standard action video is consistent with the speed of the preset martial arts movement in the target action video.

[0024] The present invention also provides a martial arts action scoring system based on artificial intelligence, the system is used to implement the above method, the system comprises:

[0025] A camera unit, used for capturing human body images of a target object at multiple angles within a first preset time, and extracting the target human body features based on the human body images;

[0026] A storage unit, configured to store a plurality of standard human features, and compare the target human feature with the plurality of standard human features to obtain a first human feature and a second human feature that are closest to the target human feature, and obtain a first standard action video corresponding to the first human feature and a second standard action video corresponding to the second human feature based on the first human feature and the second human feature;

[0027] The scoring unit is configured as follows: using the camera unit to shoot real-time action videos of the target object at N angles in real time within a second preset time, and generating a three-dimensional target action video based on the real-time action video, placing the target action video, the first standard action video and the second standard action video into the same three-dimensional space, and comparing the target action video with the first standard action video and the second standard action video, and obtaining a trajectory comparison result, and obtaining a first score based on the trajectory comparison result; also obtaining a sub-action score of each sub-action according to the completion time of each sub-action of the preset martial arts action in the target action video and the standard completion time of the sub-action based on the time difference between the completion time and the standard completion time, and obtaining a second score of the preset martial arts action according to each sub-action score; and obtaining a final score based on the first score and the second score.

[0028] The present invention also provides a computer storage medium, wherein the storage medium stores program instructions, wherein when the program instructions are executed, the device where the storage medium is located is controlled to execute the above method.

[0029] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0030] The present invention obtains more accurate target human features of the target object by shooting human images of the target object at multiple angles. Since people, i.e., objects, of different heights and body shapes have different movement trajectories when performing the same martial arts action, it is not possible to compare the same standard action video with the target action video of the target object to determine whether the preset martial arts action corresponding to the target object is standard. Moreover, since each person is unique and no one has exactly the same body shape, the first human feature and the second human feature that are closest to the target human feature are obtained. Since height is the main factor affecting the trajectory of martial arts actions, one of the first object corresponding to the first human feature and the second object corresponding to the second human feature has a height greater than the height of the target object and the other has a height less than the height of the target object, and the first human feature and the second human feature are obtained. The first standard action video corresponding to a human body feature, the second standard action video corresponding to a second human body feature and the target action video are placed in the same three-dimensional space, and the centers of gravity of the first object corresponding to the first standard action video, the second object corresponding to the second standard action video and the target object in the target action video are on a vertical line, and the feet of the three are on a horizontal line, so that the first action trajectory, the second action trajectory and the target action trajectory corresponding to the preset martial arts action respectively have the same comparison reference, and because the target human body feature of the target object is between the first human body feature and the second human body feature and the similarity of the target human body feature to the first human body feature and the second human body feature is high, the first score of the target object corresponding to the preset martial arts action can be accurately obtained by comparing the three trajectories. When evaluating martial arts actions, not only the standard degree of the action should be considered, but also the fluency of the action should be considered. Therefore, by weighting the first score and the second score to obtain the final relatively objective and accurate final score, the accuracy of the preset martial arts action score can be improved through the mutual cooperation of the above technical solutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a flow chart of a martial arts action scoring method based on artificial intelligence of the present invention;

[0032] Figure 2 The present invention is a structural diagram of a martial arts action scoring system based on artificial intelligence. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0034] The present invention provides a martial arts action scoring method based on artificial intelligence, such as Figure 1 As shown, the method comprises the following steps:

[0035] Step S1: photographing human body images of a target object at multiple angles within a first preset time by a camera unit, and extracting the target human body features based on the human body images;

[0036] Specifically, multiple cameras of the camera unit capture body images of the target object at multiple angles within the first preset time, and the first preset time is before performing martial arts movements. The technical solution is used to obtain more comprehensive body images, providing data basis for obtaining more accurate target body features.

[0037] Step S2: Compare the target human body feature with a plurality of standard human body features stored in a storage unit, obtain a first human body feature and a second human body feature that are closest to the target human body feature, and obtain a first standard action video corresponding to the first human body feature and a second standard action video corresponding to the second human body feature based on the first human body feature and the second human body feature;

[0038] Specifically, since people, i.e., objects, of different heights and body shapes have different movement trajectories when performing the same martial arts action, it is not possible to use the same standard action video and the target action video of the target object to compare to determine whether the preset martial arts action corresponding to the target object is standard, wherein the target action video is a three-dimensional video of the target object showing the preset martial arts action. Moreover, since each person is unique and no one has exactly the same body shape, the first human feature and the second human feature closest to the target human feature are obtained by comparing the target human feature with the multiple standard human features stored in the storage unit. Moreover, since height is the main factor affecting the trajectory of martial arts actions, the height of one of the first object corresponding to the first human feature and the second object corresponding to the second human feature is greater than the height of the target object and the height of the other is less than the height of the target object. By comparing the first standard action video corresponding to the first human feature and the second standard action video corresponding to the second human feature with the target action video, the target action video of the target object can be evaluated more accurately, thereby obtaining a more accurate first score.

[0039] Step S3: using the camera unit to capture real-time action videos of the target object at N angles in real time within a second preset time, and generating a three-dimensional target action video based on the real-time action video, placing the target action video, the first standard action video, and the second standard action video in the same three-dimensional space, and comparing the target action video with the first standard action video and the second standard action video, and obtaining a trajectory comparison result, and obtaining a first score based on the trajectory comparison result;

[0040] Specifically, multiple cameras of the camera unit are used to shoot the target object in real time from multiple angles within the second preset time, and obtain the real-time action videos from multiple angles, the second preset time being the time period for the target object to display the preset martial arts movements, and a three-dimensional target action video is generated based on the real-time action videos from multiple angles, and the first standard action video, the second standard action video and the target action video are placed in the same three-dimensional space, and the center of gravity of the first object corresponding to the first standard action video, the second object corresponding to the second standard action video and the target object in the target action video are on a vertical line, and the feet of the three are on a horizontal line, so that the first action trajectory, the second action trajectory and the target action trajectory corresponding to the preset martial arts movements in the first standard action video, the second standard action video and the target action video respectively have the same comparison reference. Since the target human body feature of the target object is between the first human body feature and the second human body feature and the similarity between the target human body feature and the first human body feature and the second human body feature is relatively high, when the target object displays the preset martial arts action, the target action trajectory of the target object all falls in the area between the first action trajectory corresponding to the first object and the second action trajectory corresponding to the second object. It is considered that the deviation between the target action video and the first standard action and the second standard action is small. At this time, the first score of the target object corresponding to the preset martial arts action is full marks. When the target action trajectory does not fall entirely between the first action trajectory and the second action trajectory, the first score is calculated by calculating the deviation value between the target sub-score action trajectory and the first sub-action trajectory and the second sub-action trajectory. Through the above technical solution, the first score of the target object corresponding to the preset martial arts action can be accurately obtained.

[0041] Step S4: according to the completion time of each sub-scoring action of the preset martial arts action in the target action video and the standard completion time of the sub-scoring action, based on the time difference between the completion time and the standard completion time, obtain the sub-action score of each sub-scoring action, and obtain the second score of the preset martial arts action according to each sub-action score;

[0042] Specifically, when evaluating martial arts movements, not only the standard degree of the movement should be considered, but also the fluency of the movement. Therefore, by calculating the completion time of each of the above-mentioned sub-scoring movements, and obtaining the time difference between the above-mentioned completion time and the above-mentioned sub-movement standard completion time, and obtaining the above-mentioned second score based on the above-mentioned time difference through the scoring standard, the above-mentioned second score that reflects the fluency of the above-mentioned preset martial arts movements can be obtained through the above-mentioned technical solution.

[0043] Step S5: Obtain a final score based on the first score and the second score.

[0044] Specifically, the completion status of the preset martial arts movements can be fully reflected through the first score and the second score. Therefore, the final score that is relatively objective and accurate can be obtained by weighting the first score and the second score.

[0045] Furthermore, the target human body features include the body shape of the target object, and the body shape includes height, fatness, and the proportion of each body part to the entire body. The first standard video and the second standard video are three-dimensional videos.

[0046] Specifically, since subjects with different body characteristics have different movement trajectories when performing the same martial arts move, by selecting height, fatness and the proportion of each body part to the entire body, the above-mentioned body parts refer to the proportion of the length of the limbs to the entire height. Through the above-mentioned technical solution, it is possible to obtain a more comprehensive picture of all factors affecting the above-mentioned preset martial arts moves. Since the above-mentioned first standard video and the above-mentioned second standard video are both three-dimensional videos, it is possible to obtain in all directions the deviation between the target action trajectory corresponding to the target action video and the first action trajectory corresponding to the above-mentioned first standard video and the second action trajectory corresponding to the above-mentioned second standard video, so that a more accurate first score can be obtained based on the above-mentioned target action video.

[0047] Furthermore, the step S2 comprises the following steps:

[0048] Step S21: reading a plurality of the standard human features from a storage unit, and comparing the target human feature with the height, fatness, and the proportion of each body part to the whole body in each of the standard human features, and respectively obtaining a first comparison result, a second comparison result, and a target comparison result, and weighting the first comparison result, the second comparison result, and the target comparison result to obtain a final comparison result, wherein the first comparison result has the highest weight, the target comparison result has the second highest weight, and the second comparison result has the lowest weight, and obtaining the final comparison results corresponding to all the standard human features, and taking the standard human features corresponding to the two smallest final comparison results as the first human feature and the second human feature;

[0049] Specifically, since objects with different human features have different movement trajectories when performing the same martial arts action, it is not possible to use the same standard action video and the target action video of the target object to compare to determine whether the preset martial arts action corresponding to the above target object is standard, wherein the above target action video is a three-dimensional video of the above target object displaying the preset martial arts action. Since everyone is unique and no one has exactly the same body shape, the above target human feature is compared with multiple above standard human features, wherein the larger the above final comparison result, the higher the similarity between the above target human feature and the corresponding standard human feature, and the above first human feature and the above second human feature closest to the above target human feature are obtained. Since the above height, the proportion of each body part to the whole body, and the degree of influence of fatness and thinness on the movement trajectory decrease in sequence, the weights of the three decrease in sequence, thereby accurately obtaining the first human feature and the second human feature closest to the above target human feature.

[0050] Step S22: reading the first standard action video corresponding to the first human feature and the second standard action video corresponding to the second human feature from the storage unit;

[0051] Among them, the height in any one of the first human feature and the second human feature is higher than the height in the target human feature, and the height in the other one is lower than the height in the target human feature.

[0052] Specifically, since height is the main factor affecting the trajectory of martial arts movements, the height of one of the first object corresponding to the first human feature and the second object corresponding to the second human feature is greater than the height of the target object, and the other is less than the height of the target object, so as to provide a reference for comparing the first standard action video corresponding to the first human feature and the second standard action video corresponding to the second human feature with the target action video, thereby obtaining a more accurate first score.

[0053] Furthermore, step S3 includes the following steps:

[0054] Step S31: using the multiple cameras of the camera unit to capture real-time action videos of the target object at N angles in real time within the second preset time, where N is a positive integer greater than or equal to 3;

[0055] Specifically, the target object is photographed in real time at multiple angles by the multiple cameras of the camera unit within the second preset time, and the real-time action video at multiple angles is obtained. The second preset time is the time period for the target object to perform the preset martial arts movements. The technical solution can provide rich and all-round perspective videos for further obtaining three-dimensional target action videos through the real-time action videos at multiple angles.

[0056] Step S32: generating the target action video according to the real-time action videos at N angles, and placing the target action video, the first standard action video and the second standard action video in the same three-dimensional space, wherein the center of gravity of the target object in the target action video is in a straight line with the center of gravity of the first object corresponding to the first standard action video and the second object in the second standard action video, and the feet of the target object are in the same horizontal line with the feet of the first object and the second object;

[0057] Specifically, a three-dimensional target action video is generated based on the real-time action video of the multi-angles, and the first standard action video, the second standard action video and the target action video are placed in the same three-dimensional space, and the centers of gravity of the first object corresponding to the first standard action video, the second object corresponding to the second standard action video and the target object in the target action video are on a vertical line, and the feet of the three are on a horizontal line, so that the first action trajectory, the second action trajectory and the target action trajectory corresponding to the preset martial arts actions in the first standard action video, the second standard action video and the target action video respectively have the same comparison reference, thereby laying a foundation for obtaining an accurate first score.

[0058] Step S33: Based on the first standard action video and the second standard action video, respectively obtain the first action trajectory and the second action trajectory of the preset martial arts action, obtain the target action trajectory of the preset martial arts action through the target action video, compare the target action trajectory with the first action trajectory and the second action trajectory, and obtain the trajectory comparison result. When the trajectory comparison result is that the target action trajectory falls entirely between the first action trajectory and the second action trajectory, the first score is full marks.

[0059] Specifically, since the target human feature of the target object is between the first human feature and the second human feature and the similarity between the target human feature and the first human feature and the second human feature is relatively high, therefore, when the target action trajectory of the target object when demonstrating the preset martial arts movements falls between the first action trajectory corresponding to the first object and the second action trajectory corresponding to the second object, it is considered that the deviation between the target action video and the first standard action and the second standard action is small. At this time, the first score of the target object corresponding to the preset martial arts movements is full marks. Through the above technical solution, the first score can be accurately obtained when the target action trajectory corresponding to the target object falls between the first action trajectory and the second action trajectory.

[0060] Further, when the first comparison result is that the target motion trajectory is not all in the area between the first motion trajectory and the second motion trajectory, the target motion trajectory is divided into a plurality of target sub-scoring motion trajectories, and the first motion trajectory and the second motion trajectory are respectively divided into a plurality of first sub-scoring motion trajectories and a plurality of second sub-scoring motion trajectories, wherein each of the target sub-scoring motion trajectories corresponds to each of the first sub-scoring motion trajectory and the second sub-scoring motion trajectory, respectively, and when the target sub-scoring motion trajectory falls between the first sub-scoring motion trajectory and the second sub-scoring motion trajectory, the sub-motion scoring corresponding to the target sub-scoring motion trajectory is full marks; when the target sub-scoring action trajectory falls on both sides of the first sub-scoring action trajectory and the second sub-scoring action trajectory, calculate the deviation value between the target sub-scoring action trajectory and the first sub-scoring action trajectory or the second sub-scoring action trajectory which is closer to the target sub-scoring action trajectory, and when the deviation value is less than or equal to a first threshold value, the sub-action score corresponding to the target sub-scoring action trajectory is full marks; when the deviation value is greater than the first threshold value, obtain the sub-action score according to the deviation value, wherein the larger the deviation value is, the smaller the sub-action score is, and vice versa, the larger the sub-action score is, and the first score is obtained based on each sub-action score.

[0061] Specifically, when the target action trajectory does not entirely fall between the first action trajectory and the second action trajectory, by dividing the target action trajectory into a plurality of target sub-scoring action trajectories, the first action trajectory and the second action trajectory are respectively divided into a plurality of first sub-scoring action trajectories and a plurality of second sub-scoring action trajectories, and when the target sub-scoring action trajectory falls between the first sub-scoring action trajectory and the second sub-scoring action trajectory, the sub-action score corresponding to the target sub-scoring action trajectory is full marks, and when the target sub-scoring action trajectory falls on both sides of the first sub-scoring action trajectory and the second sub-scoring action trajectory, the outer side here refers to the side of the first sub-action trajectory away from the second sub-action trajectory and the side of the second sub-action trajectory away from the first sub-action trajectory. When scoring, since the target sub-scoring action trajectory is close to any one of the first sub-action trajectory and the second sub-action trajectory, the sub-action is considered to be standard. Therefore, when the target sub-scoring action trajectory falls on both sides of the first sub-action trajectory and the second sub-action trajectory, the sub-action score is obtained by calculating the deviation value. When the deviation value is less than or equal to the first threshold value, the sub-action is considered to be standard. When the deviation value is greater than the first threshold value, the sub-action is considered not to be standard. Therefore, the sub-action score of the sub-action is obtained through the deviation value and the scoring table corresponding to the deviation value, and the first score is obtained by the sum of the sub-action scores of each sub-action. Through the above technical solution, the first score of the target object corresponding to the preset martial arts action can be accurately obtained.

[0062] Furthermore, the step S4 comprises the following steps:

[0063] Step S41: dividing the preset martial arts action into a plurality of sub-scoring actions, and respectively calculating the completion time of each of the sub-scoring actions in the target action video;

[0064] Step S42: Compare the completion time with the standard completion time of the sub-scoring action to obtain the time difference, and obtain the sub-action score of the sub-scoring action based on the time difference and the scoring standard, and obtain the second score of the preset martial arts action based on each sub-action score.

[0065] Specifically, when evaluating martial arts movements, not only the standard degree of the movement should be considered, but also the fluency of the movement. Therefore, by calculating the completion time of each of the above-mentioned sub-scoring movements, and obtaining the time difference between the above-mentioned completion time and the above-mentioned sub-movement standard completion time, and obtaining the above-mentioned sub-movement score based on the above-mentioned time difference through the scoring standard, the above-mentioned second score reflecting the fluency of the above-mentioned preset martial arts movement can be obtained through the above-mentioned technical solution.

[0066] Further, the step S5 includes: calculating the final score according to the weight of the first score and the weight of the second score, wherein the first score weight is greater than the second score weight.

[0067] Specifically, the completion status of the preset martial arts movements can be fully reflected through the first score and the second score. Therefore, the final score that is relatively objective and accurate can be obtained by weighting the first score and the second score.

[0068] Furthermore, the step S3 also includes, when the speed of the preset martial arts movement in the target action video is inconsistent with the speed of the preset martial arts movement in the first standard action video and the second standard action video, adjusting the action speed of the first standard action video and the second standard action video so that the speed of the preset martial arts movement in the first standard action video and the second standard action video is consistent with the speed of the preset martial arts movement in the target action video.

[0069] Specifically, in order to be able to timely score each sub-action in the above-mentioned target action video, by adjusting the above-mentioned preset martial arts actions in the above-mentioned first standard video and the above-mentioned second standard video to be consistent in speed with the above-mentioned preset martial arts actions in the above-mentioned target action video, the above-mentioned target action trajectory, the above-mentioned first action trajectory and the above-mentioned second action trajectory can be obtained at the same time, and then each sub-action in the above-mentioned preset martial arts actions can be scored in time, thereby improving the scoring efficiency.

[0070] The present invention also provides a martial arts action scoring system based on artificial intelligence, the system is used to implement the above method, such as Figure 2 As shown, the system comprises:

[0071] A camera unit, used for capturing human body images of a target object at multiple angles within a first preset time, and extracting the target human body features based on the human body images;

[0072] A storage unit, configured to store a plurality of standard human features, and compare the target human feature with the plurality of standard human features to obtain a first human feature and a second human feature that are closest to the target human feature, and obtain a first standard action video corresponding to the first human feature and a second standard action video corresponding to the second human feature based on the first human feature and the second human feature;

[0073] The scoring unit is configured as follows: using the camera unit to shoot real-time action videos of the target object at N angles in real time within a second preset time, and generating a three-dimensional target action video based on the real-time action video, placing the target action video, the first standard action video and the second standard action video into the same three-dimensional space, and comparing the target action video with the first standard action video and the second standard action video, and obtaining a trajectory comparison result, and obtaining a first score based on the trajectory comparison result; also obtaining a sub-action score of each sub-action according to the completion time of each sub-action of the preset martial arts action in the target action video and the standard completion time of the sub-action based on the time difference between the completion time and the standard completion time, and obtaining a second score of the preset martial arts action according to each sub-action score; and obtaining a final score based on the first score and the second score.

[0074] The present invention also provides a computer storage medium, wherein the storage medium stores program instructions, wherein when the program instructions are executed, the device where the storage medium is located is controlled to execute the above method.

[0075] In summary, the present invention obtains more accurate target human features of the target object by shooting human images of the target object at multiple angles. Since people, i.e., objects, of different heights and body shapes have different movement trajectories when performing the same martial arts action, it is not possible to use the same standard action video and the target action video of the target object for comparison to determine whether the preset martial arts action corresponding to the target object is standard. Moreover, since each person is unique and there are no people with exactly the same body shapes, the first human feature and the second human feature that are closest to the target human feature are obtained. Since height is the main factor affecting the trajectory of martial arts actions, one of the first object corresponding to the first human feature and the second object corresponding to the second human feature has a height greater than the height of the target object and the other has a height less than the height of the target object, and The first standard action video corresponding to the first human feature, the second standard action video corresponding to the second human feature and the target action video are placed in the same three-dimensional space, and the centers of gravity of the first object corresponding to the first standard action video, the second object corresponding to the second standard action video and the target object in the target action video are on a vertical line, and the feet of the three are on a horizontal line, so that the first action trajectory, the second action trajectory and the target action trajectory of the three corresponding to the preset martial arts action have the same comparison reference, and because the target human feature of the target object is between the first human feature and the second human feature and the similarity of the target human feature to the first human feature and the second human feature is high, the first score of the target object corresponding to the preset martial arts action can be accurately obtained by comparing the three trajectories. When evaluating martial arts actions, not only the standard degree of the action should be considered, but also the fluency of the action should be considered. Therefore, by weighting the first score and the second score to obtain the final relatively objective and accurate final score, the accuracy of the preset martial arts action score can be improved through the mutual cooperation of the above technical solutions.

[0076] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0077] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

[0078] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A martial arts action scoring method based on artificial intelligence, It is characterized in that The method comprises the following steps: Step S1: photographing human body images of a target object at multiple angles within a first preset time by a camera unit, and extracting target human body features based on the human body images; Step S2: Compare the target human body feature with a plurality of standard human body features stored in a storage unit, obtain a first human body feature and a second human body feature that are closest to the target human body feature, and obtain a first standard action video corresponding to the first human body feature and a second standard action video corresponding to the second human body feature based on the first human body feature and the second human body feature; Step S3: using the camera unit to capture real-time action videos of the target object at N angles in real time within a second preset time, and generating a three-dimensional target action video based on the real-time action video, placing the target action video, the first standard action video, and the second standard action video in the same three-dimensional space, and comparing the target action video with the first standard action video and the second standard action video, and obtaining a trajectory comparison result, and obtaining a first score based on the trajectory comparison result; Step S4: according to the completion time of each sub-scoring action of the preset martial arts action in the target action video and the standard completion time of the sub-scoring action, based on the time difference between the completion time and the standard completion time, obtain the sub-action score of each sub-scoring action, and obtain the second score of the preset martial arts action according to each sub-action score; Step S5: obtaining a final score based on the first score and the second score; The first preset time is a time period before the real-time action video of the target object is captured, and the second preset time is a time period when the target object performs the preset martial arts action. Among them, the height in any one of the first human feature and the second human feature is higher than the height in the target human feature, and the height in the other one is lower than the height in the target human feature.

2. The method according to claim 1, It is characterized in that The target human body features include the body shape of the target object, and the body shape includes height, fatness, and the proportion of each body part to the entire body. The first standard action video and the second standard action video are three-dimensional videos.

3. The method according to claim 2, It is characterized in that The step S2 comprises the following steps: Step S21: reading a plurality of the standard human features from a storage unit, and comparing the target human feature with the height, fatness, and proportion of each body part in the whole body in each of the standard human features, and respectively obtaining a first comparison result, a second comparison result, and a third comparison result, and weighting the first comparison result, the second comparison result, and the third comparison result to obtain a final comparison result, wherein the first comparison result has the highest weight, the third comparison result has the second highest weight, and the second comparison result has the lowest weight, and obtaining the final comparison results corresponding to all the standard human features, and taking the standard human features corresponding to the two smallest final comparison results as the first human feature and the second human feature; Step S22: Reading the first standard action video corresponding to the first human feature and the second standard action video corresponding to the second human feature from the storage unit.

4. The method according to claim 1, It is characterized in that The step S3 comprises the following steps: Step S31: using the multiple cameras of the camera unit to capture real-time action videos of the target object at N angles in real time within the second preset time, where N is a positive integer greater than or equal to 3; Step S32: generating the target action video according to the real-time action videos at N angles, and placing the target action video, the first standard action video and the second standard action video in the same three-dimensional space, wherein the center of gravity of the target object in the target action video is in a straight line with the center of gravity of the first object corresponding to the first standard action video and the second object in the second standard action video, and the feet of the target object are in the same horizontal line with the feet of the first object and the second object; Step S33: Based on the first standard action video and the second standard action video, respectively obtain the first action trajectory and the second action trajectory of the preset martial arts action, obtain the target action trajectory of the preset martial arts action through the target action video, compare the target action trajectory with the first action trajectory and the second action trajectory, and obtain the trajectory comparison result. When the trajectory comparison result is that the target action trajectory falls entirely between the first action trajectory and the second action trajectory, the first score is full marks.

5. The method according to claim 4, It is characterized in that When the trajectory comparison result shows that the target motion trajectory does not completely fall between the first motion trajectory and the second motion trajectory, the target motion trajectory is divided into a plurality of target sub-scoring motion trajectories, and the first motion trajectory and the second motion trajectory are respectively divided into a plurality of first sub-scoring motion trajectories and a plurality of second sub-scoring motion trajectories, wherein each of the target sub-scoring motion trajectories corresponds to each of the first sub-scoring motion trajectory and the second sub-scoring motion trajectory, respectively, and when the target sub-scoring motion trajectory falls between the first sub-scoring motion trajectory and the second sub-scoring motion trajectory, the sub-motion score corresponding to the target sub-scoring motion trajectory is full score ; When the target sub-scoring action trajectory falls on both sides of the first sub-scoring action trajectory and the second sub-scoring action trajectory, calculate the deviation value between the target sub-scoring action trajectory and the first sub-scoring action trajectory or the second sub-scoring action trajectory which is closer to the target sub-scoring action trajectory, and when the deviation value is less than or equal to a first threshold, the sub-action score corresponding to the target sub-scoring action trajectory is full marks; when the deviation value is greater than the first threshold, obtain the sub-action score according to the deviation value, wherein the larger the deviation value is, the smaller the sub-action score is, and vice versa, the larger the sub-action score is, and obtain the first score based on each sub-action score.

6. The method according to claim 1, It is characterized in that The step S4 comprises the following steps: Step S41: dividing the preset martial arts action into a plurality of sub-scoring actions, and respectively calculating the completion time of each of the sub-scoring actions in the target action video; Step S42: Compare the completion time with the standard completion time of the sub-scoring action to obtain the time difference, and obtain the sub-action score of the sub-scoring action based on the time difference and the scoring standard, and obtain the second score of the preset martial arts action based on each sub-action score.

7. The method according to claim 1, It is characterized in that The step S5 includes: calculating the final score according to the weight of the first score and the weight of the second score, wherein the weight of the first score is greater than the weight of the second score.

8. The method according to claim 1, It is characterized in that The step S3 also includes, when the speed of the preset martial arts movement in the target action video is inconsistent with the speed of the preset martial arts movement in the first standard action video and the second standard action video, adjusting the action speed of the first standard action video and the second standard action video so that the speed of the preset martial arts movement in the first standard action video and the second standard action video is consistent with the speed of the preset martial arts movement in the target action video.

9. A martial arts action scoring system based on artificial intelligence, the system being used to implement the method according to any one of claims 1 to 8, It is characterized in that The system comprises: A camera unit, used for capturing human body images of a target object at multiple angles within a first preset time, and extracting the target human body features based on the human body images; A storage unit, configured to store a plurality of standard human features, and compare the target human feature with the plurality of standard human features to obtain a first human feature and a second human feature that are closest to the target human feature, and obtain a first standard action video corresponding to the first human feature and a second standard action video corresponding to the second human feature based on the first human feature and the second human feature; The scoring unit is configured as follows: using the camera unit to shoot real-time action videos of the target object at N angles in real time within a second preset time, and generating a three-dimensional target action video based on the real-time action video, placing the target action video, the first standard action video and the second standard action video into the same three-dimensional space, and comparing the target action video with the first standard action video and the second standard action video, and obtaining a trajectory comparison result, and obtaining a first score based on the trajectory comparison result; also obtaining a sub-action score of each sub-action according to the completion time of each sub-action of the preset martial arts action in the target action video and the standard completion time of the sub-action based on the time difference between the completion time and the standard completion time, and obtaining a second score of the preset martial arts action according to each sub-action score; and obtaining a final score based on the first score and the second score.

10. A computer storage medium, It is characterized in that The storage medium stores program instructions, wherein when the program instructions are executed, the device where the storage medium is located is controlled to execute the method according to any one of claims 1 to 8.