Vision-based take-off violation detection method for standing long jump
By adding key point detection of the toes and multi-dimensional violation detection to the standing long jump test, and combining computer vision and posture recognition algorithms, the problem of inaccurate violation identification in traditional detection is solved, realizing objective and accurate detection of take-off violations, and ensuring the standardization and fairness of the test.
Patent Information
- Application Number
- CN202610122242.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-29
- Publication Date
- 2026-05-12
AI Technical Summary
Existing technologies for detecting violations in the standing long jump take-off lack the ability to accurately identify subtle violations. Traditional human body key point models are unable to detect key parts such as the toes, and there is a lack of a comprehensive scoring mechanism that considers multiple violation judgment conditions, resulting in unstable and inaccurate detection results.
A vision-based method for detecting violations in the standing long jump takeoff is adopted. By adding key points for the left and right toes to the traditional 17-point human body model, and combining computer vision and posture recognition algorithms, the coordinates of 19 key points are obtained. Multi-dimensional violation indicators are comprehensively calculated, and a quantitative judgment model is established to achieve objective and accurate detection of takeoff violations.
It enables accurate identification and automatic warning of violations during the take-off of the standing long jump, eliminates human observation errors, ensures the standardization and fairness of the test, and improves the accuracy and consistency of the test.
Smart Images

Figure CN122015777A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of physical fitness testing technology, specifically to a vision-based method for detecting violations during the standing long jump take-off. Background Technology
[0002] The standing long jump is an important event in physical education and physical fitness testing, and the standardization of the take-off action directly affects the accuracy and fairness of the test results. Traditional standing long jump testing mainly relies on manual observation and subjective judgment to identify take-off violations. However, with the continuous improvement of testing standardization requirements, intelligent detection methods based on computer vision technology have gradually become a research hotspot.
[0003] In the existing technology, there are several standing long jump testing schemes based on vision technology. Chinese patent CN113137923A discloses a method for measuring standing long jump performance. This method uses a camera and machine vision technology to achieve intelligent measurement of standing long jump performance, including steps such as setting the test area and motion parameters, collecting user test videos, and analyzing user test videos to obtain scores. Chinese patent CN114712769A proposes a computer vision-based intelligent distance measurement method and system for standing long jump. By combining computer vision with human key point recognition technology, it photographs the landing point on the jumping mat during the standing long jump and uses human key point recognition to ultimately confirm the landing point and calculate the long jump score. Chinese patent CN110624203B discloses a method and device for freestyle standing long jump testing. It uses a camera and machine vision to perform rule judgment and score measurement for freestyle standing long jump testing, overcoming the limitation of traditional standing long jump testing relying on a take-off line. Chinese patent CN117379769A discloses a computer vision-based method for analyzing and measuring the distance in a standing long jump. Based on the temporal movement localization of key points in the human skeleton, it can analyze and measure the entire standing long jump movement, both identifying fouls and automatically measuring the distance. Chinese patent CN117582649B provides a computer vision-based dual-camera three-level frog jump testing method and system. It uses two cameras to monitor the preparation area and the long jump area respectively, and can determine in real time whether the take-off is illegal or the jump is discontinuous.
[0004] However, existing technologies still have the following shortcomings in detecting violations during the standing long jump takeoff: First, existing solutions mainly focus on performance measurement and landing point detection, and are not precise enough in detecting violations during the takeoff phase, lacking the ability to accurately identify subtle violations such as foot movements and hand touches; second, existing human key point recognition models mostly use the traditional 17-point model, lacking detection points for key parts such as the toes, making it difficult to accurately judge violations such as stepping on the foot line; third, existing technologies lack a scoring mechanism that integrates multiple violation judgment conditions, and cannot provide stable and reliable violation detection results in complex testing environments, affecting the practicality and accuracy of the detection system. Summary of the Invention
[0005] To address the limitations, subjectivity, and inconsistent standards inherent in traditional standing long jump start violation detection methods that rely primarily on teacher observation and subjective judgment, this invention provides a vision-based method for detecting start violations in the standing long jump.
[0006] The objective of this invention is achieved through the following technical solution: a vision-based method for detecting violations during the standing long jump takeoff, comprising the following steps: 1) Set up a camera on the side of the standing long jump test area to capture video images containing the take-off area, and mark the coordinates of the four vertices of the take-off area rectangle as the left vertex (x1, y1), the right vertex (x2, y1), the left bottom corner (x1, y2), and the right bottom corner (x2, y2), and the coordinates of the two edge points of the take-off line as the edge point closer to the camera (x2, y3) and the edge point farther from the camera (x2, y4). 2) Based on the human key point recognition model, the coordinates of 19 key points of the tester during the take-off process were obtained, including the coordinates of the left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip, right hip, left knee, right knee, left foot, right foot, left toe, and right toe. 3) Based on the shoulder coordinates (slx) of key points in each frame of the image i sly i ), right shoulder coordinates (srx) i ,sry i ), left elbow coordinates (elx) i ,ely i Right elbow coordinates (erx) i ery i ), left wrist coordinates (hlx) i ,hly i ), right wrist coordinates (hrx) i hry i ), left hip coordinates (ulx) i uly i ), right hip coordinate urx i ,ury i Left knee coordinates (klx) i ,kly i ), right knee coordinates (krx) i ,kry i ), left foot coordinates (flx) i fly i ), right foot coordinate (frx) i fry i ), left toe coordinates (blx)i bly i ), right toe coordinates (brx) i ,bry i The test determines whether the test subject's feet are within the rectangular take-off area. If the horizontal and vertical coordinates of the key points of both feet are within the take-off area, then the test subject is considered to have feet within the area. 4) Determine whether the test subject has taken off. If the x-coordinates of all key points of the lower limbs and torso are greater than the x-coordinate of the take-off line, then it is determined that the subject has taken off. 5) Determine whether the tester moved their feet before jumping by calculating the displacement distance of the key points of the toes and ankles relative to the initial frame and comparing it with a set threshold. If the displacement is less than the threshold, it is determined that there was no foot movement. 6) Determine whether the tester's hands are in contact with the ground in front of the take-off line. Calculate the torso posture angle and the geometric relationship between the wrist and the ground. If the set conditions are met, it is determined that the hands are in contact with the ground. 7) Starting from the n1th frame after the system issues the start command, based on the judgment results of the aforementioned steps, the jump violation score is calculated comprehensively. If the conditions of both feet being in the area, no foot movement, and no hand touching the ground are met within a certain time window, the jump is judged to be compliant; otherwise, it is judged to be a violation, and a violation warning is output.
[0007] Preferably, the 19 key point recognition model in step 2) is based on the traditional 17-point human body model under the YOLO framework, with the addition of two key points, left toe and right toe, and is trained using manually annotated image samples of the take-off phase.
[0008] Preferably, in step 3), when the system issues the start command, the positions of key points of the person within the take-off area rectangle of each frame i are obtained; when determining whether the feet are within the take-off area rectangle, the foot position score g1 is calculated. i To determine: , Among them, g1 i =1 indicates that the person's feet are within the detection area; otherwise, it is determined that the feet are outside the boundary.
[0009] Preferably, in step 4), when determining whether a person has taken off, the takeoff score g2 is calculated. i To determine: , Among them, g2 i =1 indicates that the person has jumped; otherwise, it is considered that they have not jumped.
[0010] Preferably, in step 5), when determining whether the tester moved their feet before taking off, the foot movement score g3 is calculated. i To determine: , Wherein, ts1 is the set first judgment threshold, which is calculated by collecting the maximum value of the range of fluctuation distance of the toe coordinate before the long jump take-off; ts2 is the set second judgment threshold, which is calculated by collecting the maximum value of the range of fluctuation distance of the ankle coordinate before the long jump take-off. g3 i =1 indicates that the person's feet have not moved; otherwise, it is determined that there has been foot movement.
[0011] Preferably, in step 6), when determining whether the tester's hands have touched the ground in front of the takeoff line, the score g4 for hands touching the ground is calculated. i To determine: , Among them, g41 i This indicates a point scored by touching the ground with the left hand, g42 i Indicates a point scored by touching the ground with the right hand: , gx41 i Let gx42 be the left-hand horizontal coordinate of the landing point. i The right-hand landing x-coordinate: , ts3 is the set third judgment threshold, which is calculated by collecting ground images before the contact line in the long jump and obtaining the cosine value corresponding to the maximum value of the angle between the torso and the vertical ground normal. ts4 is the fourth judgment threshold, which is calculated by collecting images of the ground before the contact line in the long jump and obtaining the maximum value of the ratio of the distance from the fingertip to the wrist to the distance from the wrist to the elbow.
[0012] Preferably, in step 7), when determining whether the tester violates the rules for the pre-jump long jump starting from frame n1, the violation score g5 is calculated. i To determine: , ts5 is the fifth judgment threshold set, which is obtained by collecting the minimum value of the proportion of illegal action frames in each frame of the illegal jump video within the corresponding n1 frame time window, and n1 is a positive integer calculated based on the minimum value of consecutive illegal action frames when an illegal action occurs. When g5 i When =1, the jumper is deemed to have violated the rules; otherwise, g5 i =0, deemed compliant.
[0013] Preferably, when g5 iWhen =1, it is determined that the person's jump is in violation of the rules. When a violation occurs, the system immediately sends a violation warning to the test terminal through sound and light or wireless message, and automatically resets to the waiting jump state.
[0014] Preferably, all thresholds ts1-ts5 frame count n1 can be dynamically configured by the administrator based on the venue, camera resolution and test level during the initialization of the test system, and stored in a local configuration file or remote database.
[0015] Preferably, the camera is installed at a height of 1.0-1.5m, and the vertical distance between the optical axis and the jump line is 0.8-1.2m, ensuring that the rectangle of the jump area occupies ≥60% of the image.
[0016] The beneficial effects of this invention are as follows: by using computer vision and posture recognition algorithms, key parameters such as the force exertion posture of students when jumping are accurately identified, and a quantitative judgment model for violations is established. Compared with the traditional method that relies on teachers' on-site observation and subjective judgment, this invention achieves objective, accurate and consistent detection of jumping violations, eliminates human observation errors, ensures the standardization and fairness of test evaluation, and solves the problems of observation limitations, strong subjectivity and inconsistent standards in traditional methods. Attached Figure Description
[0017] Figure 1 This is a flowchart illustrating the testing method of the present invention; Figure 2 This is a schematic diagram of the test system of the present invention; Figure 3 This is a schematic diagram of the key points of the human skeleton in this invention.
[0018] The attached diagram is labeled as follows: 0-nose, 1-left eye, 2-right eye, 3-left ear, 4-right ear, 5-left shoulder, 6-right shoulder, 7-left elbow, 8-right elbow, 9-left wrist, 10-right wrist, 11-left hip, 12-right hip, 13-left knee, 14-right knee, 15-left ankle, 16-right ankle, 17-left toe, 18-right toe. Detailed Implementation
[0019] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to embodiments and accompanying drawings. The content mentioned in the embodiments is not intended to limit the present invention.
[0020] See Figure 1-3 A vision-based method for detecting violations during the standing long jump take-off is proposed, which uses computer vision technology and human key point recognition technology to automatically detect and warn of violations during the standing long jump take-off.
[0021] Step 1: Camera Installation and Area Calibration. Install a camera on the side of the standing long jump test area at a height of 1.0-1.5m, with the optical axis perpendicular to the take-off line at a distance of 0.8-1.2m, ensuring that the take-off area rectangle occupies more than 60% of the frame. The camera captures video images containing the take-off area. The coordinates of the four vertices of the take-off area rectangle are calibrated as follows: left vertex (x1, y1), right vertex (x2, y1), left bottom corner (x1, y2), and right bottom corner (x2, y2). Simultaneously, the coordinates of the two edge points of the take-off line are calibrated as the edge point closer to the camera (x2, y3) and the edge point farther from the camera (x2, y4).
[0022] Step 2: Construction of the Human Keypoint Recognition Model. Based on the YOLO framework, two additional keypoints, left toe and right toe, are added to the traditional 17-point human body model, forming a 19-keypoint human body recognition model. Screenshots of the take-off phase are collected from long jump videos, and the toe positions are manually marked on the images. These manually marked take-off phase image samples are used to train the YOLO model, resulting in a human keypoint recognition model capable of recognizing all 19 keypoints. This model can obtain the coordinates of the test subject's left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip, right hip, left knee, right knee, left foot, right foot, left toe, and right toe during the take-off process.
[0023] Step 3: Keypoint Coordinate Acquisition and Foot Position Determination. When the system issues the start command, it acquires the coordinates of 19 keypoints of the test subject in each frame of the image, including shoulder coordinates (SLX). i sly i ), right shoulder coordinates (srx) i ,sry i ), left elbow coordinates (elx) i ,ely i Right elbow coordinates (erx) i ery i ), left wrist coordinates (hlx) i ,hly i ), right wrist coordinates (hrx) i hry i ), left hip coordinates (ulx) i uly i ), right hip coordinate urx i ,ury i Left knee coordinates (klx) i ,kly i ), right knee coordinates (krx) i ,kry i ), left foot coordinates (flx) i fly i), right foot coordinate (frx) i fry i ), left toe coordinates (blx) i bly i ), right toe coordinates (brx) i ,bry i The score g1 is calculated by measuring the position of both feet. i To determine whether the test subject's feet are within the rectangular take-off area: , g1 i =1 indicates that the person's feet are within the detection area; otherwise, it is determined that the feet are outside the boundary.
[0024] Step 4: Determine the takeoff status. Calculate the takeoff score g2. i To determine whether the tester has jumped: , When the x-coordinates of all key points in the lower limbs and trunk are greater than the x-coordinate of the take-off line, g2 i =1 indicates that the person has jumped; otherwise, it is considered that they have not jumped.
[0025] Step 5: Foot movement detection. Calculate the foot movement score g3. i To determine whether the test subject moved their feet before taking the jump: , Where ts1 is the set first judgment threshold, calculated by collecting the maximum value of the fluctuation range of the toe coordinate before the long jump takeoff; ts2 is the set second judgment threshold, calculated by collecting the maximum value of the fluctuation range of the ankle coordinate before the long jump takeoff. When g3 i =1 indicates that the tester's feet did not move; otherwise, it is determined that there was foot movement.
[0026] Step 6: Hand-to-ground detection. The score g4 for both hands touching the ground is calculated. i To determine whether the tester's hands are in contact with the ground in front of the takeoff line: , Among them, g41 i This indicates a point scored by touching the ground with the left hand, g42 i Indicates a point scored by touching the ground with the right hand: , gx41 i Let gx42 be the left-hand horizontal coordinate of the landing point. i The right-hand landing x-coordinate: , ts3 is the set third judgment threshold, which is calculated by collecting ground images before the contact line in the long jump and obtaining the cosine value corresponding to the maximum value of the angle between the torso and the vertical ground normal. ts4 is the fourth judgment threshold, which is calculated by collecting images of the ground before the contact line in the long jump and obtaining the maximum value of the ratio of the distance from the fingertip to the wrist to the distance from the wrist to the elbow.
[0027] Step 7: Comprehensive Judgment and Early Warning Output of Violations. Starting from the n1th day after the system issues the start command, the jump violation score g5 is calculated. i To determine whether the tester's jump violated regulations: , ts5 is the fifth judgment threshold set, which is obtained by collecting the minimum value of the proportion of illegal action frames in each frame of the illegal jump video within the corresponding n1 frame time window, and n1 is a positive integer calculated based on the minimum value of consecutive illegal action frames when an illegal action occurs. When g5 i When the value is 1, the tester's jump is deemed to have violated the rules. When a violation occurs, the system immediately issues a violation warning to the test terminal via sound, light, or wireless message, and automatically resets to the waiting jump state. If, within a certain time window, both feet are within the designated area, there is no foot movement, and no hands touch the ground, the jump is deemed compliant.
[0028] In a preferred embodiment, all thresholds ts1-ts5 and frame number n1 can be dynamically configured by the administrator based on the venue, camera resolution and test level during system initialization, and stored in a local configuration file or remote database, thereby improving the system's adaptability and flexibility.
[0029] This method, through multi-dimensional violation detection indicators and a comprehensive evaluation mechanism within a time window, can accurately identify various violations during the standing long jump take-off process, achieving automated and intelligent violation detection and improving the fairness and efficiency of the test.
[0030] The above embodiments are preferred implementations of the present invention. In addition, the present invention can be implemented in other ways. Any obvious substitutions without departing from the concept of the present invention are within the protection scope of the present invention.
Claims
1. A vision-based method for detecting violations of the standing long jump take-off rule, characterized in that: Includes the following steps: 1) Set up a camera on the side of the standing long jump test area to capture video images containing the take-off area, and mark the coordinates of the four vertices of the take-off area rectangle as the left vertex (x1, y1), the right vertex (x2, y1), the left bottom corner (x1, y2), and the right bottom corner (x2, y2), and the coordinates of the two edge points of the take-off line as the edge point closer to the camera (x2, y3) and the edge point farther from the camera (x2, y4). 2) Based on the human key point recognition model, the coordinates of 19 key points of the tester during the take-off process were obtained, including the coordinates of the left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip, right hip, left knee, right knee, left foot, right foot, left toe, and right toe. 3) Based on the shoulder coordinates (slx) of key points in each frame of the image i sly i ), right shoulder coordinates (srx) i ,sry i ), left elbow coordinates (elx) i ,ely i Right elbow coordinates (erx) i ery i ), left wrist coordinates (hlx) i ,hly i ), right wrist coordinates (hrx) i hry i ), left hip coordinates (ulx) i uly i ), right hip coordinates (urx) i ,ury i Left knee coordinates (klx) i ,kly i ), right knee coordinates (krx) i ,kry i ), left foot coordinates (flx) i fly i ), right foot coordinate (frx) i fry i ), left toe coordinates (blx) i bly i ), right toe coordinates (brx) i ,bry i The test determines whether the test subject's feet are within the rectangular take-off area. If the horizontal and vertical coordinates of the key points of both feet are within the take-off area, then the test subject is considered to have feet within the area. 4) Determine whether the test subject has taken off. If the x-coordinates of all key points of the lower limbs and torso are greater than the x-coordinate of the take-off line, then it is determined that the subject has taken off. 5) Determine whether the tester moved their feet before jumping by calculating the displacement distance of the key points of the toes and ankles relative to the initial frame and comparing it with a set threshold. If the displacement is less than the threshold, it is determined that there was no foot movement. 6) Determine whether the tester's hands are in contact with the ground in front of the take-off line. Calculate the torso posture angle and the geometric relationship between the wrist and the ground. If the set conditions are met, it is determined that the hands are in contact with the ground. 7) Starting from the n1th frame after the system issues the start command, based on the judgment results of the aforementioned steps, the jump violation score is calculated comprehensively. If the conditions of both feet being in the area, no foot movement, and no hand touching the ground are met within a certain time window, the jump is judged to be compliant; otherwise, it is judged to be a violation, and a violation warning is output.
2. The vision-based standing long jump take-off violation detection method according to claim 1, characterized in that: The 19 key point recognition models in step 2) are based on the traditional 17-point human body model under the YOLO framework, with the addition of two key points, left toe and right toe, and are trained using manually annotated images of the take-off phase.
3. The vision-based standing long jump take-off violation detection method according to claim 1, characterized in that: In step 3), when the system issues the start command, the positions of key points of the personnel within the take-off area rectangle of each frame are obtained; when determining whether the feet are within the take-off area rectangle, the foot position score g1 is calculated. i To determine: , Among them, g1 i =1 indicates that the person's feet are within the detection area; otherwise, it is determined that the feet are outside the boundary.
4. The vision-based standing long jump take-off violation detection method according to claim 1, characterized in that: In step 4), when determining whether a person has taken off, the takeoff score g2 is calculated. i To determine: , Among them, g2 i =1 indicates that the person has jumped; otherwise, it is considered that they have not jumped.
5. The vision-based method for detecting violations of the standing long jump take-off rule as described in claim 1, characterized in that: In step 5), when determining whether the tester moved their feet before the jump, the foot movement score g3 is calculated. i To determine: , Wherein, ts1 is the first judgment threshold, which is calculated by collecting the maximum value of the range of fluctuation distance of the toe coordinate before the long jump take-off; ts2 is the second judgment threshold, which is calculated by collecting the maximum value of the range of fluctuation distance of the ankle coordinate before the long jump take-off. g3 i =1 indicates that the person's feet have not moved; otherwise, it is determined that there has been foot movement.
6. The vision-based standing long jump take-off violation detection method according to claim 5, characterized in that: In step 6), when determining whether the tester's hands touched the ground in front of the takeoff line, the score g4 for hand contact with the ground is calculated. i To determine: , Among them, g41 i This indicates a point scored by touching the ground with the left hand, g42 i Indicates a point scored by touching the ground with the right hand: , gx41 i Let gx42 be the left-hand horizontal coordinate of the landing point. i The right-hand landing x-coordinate: , ts3 is the set third judgment threshold, which is calculated by collecting ground images before the contact line in the long jump and obtaining the cosine value corresponding to the maximum value of the angle between the torso and the vertical ground normal. ts4 is the fourth judgment threshold, which is calculated by collecting images of the ground before the contact line in the long jump and obtaining the maximum value of the ratio of the distance from the fingertip to the wrist to the distance from the wrist to the elbow.
7. The vision-based standing long jump take-off violation detection method according to claim 6, characterized in that: In step 7), when determining whether the tester violated the rules for the pre-jump long jump starting from frame n1, the violation score g5 is calculated. i To determine: , ts5 is the fifth judgment threshold set, which is obtained by collecting the minimum value of the proportion of illegal action frames in each frame of the illegal jump video within the corresponding n1 frame time window, and n1 is a positive integer calculated based on the minimum value of consecutive illegal action frames when an illegal action occurs. When g5 i When =1, the jumper is deemed to have violated the rules; otherwise, g5 i =0, deemed compliant.
8. The vision-based standing long jump take-off violation detection method according to claim 7, characterized in that: When g5 i When =1, it is determined that the person's jump is in violation of the rules. When a violation occurs, the system immediately sends a violation warning to the test terminal through sound and light or wireless message, and automatically resets to the waiting jump state.
9. The vision-based method for detecting violations of the standing long jump take-off rule as described in claim 7, characterized in that: All thresholds ts1-ts5 and frame count n1 can be dynamically configured by the administrator during system initialization based on the venue, camera resolution, and test level, and stored in a local configuration file or remote database.
10. The vision-based standing long jump take-off violation detection method according to claim 9, characterized in that: The camera is installed at a height of 1.0-1.5m, and the vertical distance between the optical axis and the jump line is 0.8-1.2m, ensuring that the rectangular jump area occupies ≥60% of the image.
