Track and field running timing and circle recording method and system based on computer vision

Abstract

The invention provides a track and field running timing and circle recording method and system based on computer vision, and relates to the technical field of data processing. Number plates on the outer sides of thighs of athletes are recognized through a thousand-frame camera, current electronic circle recording is replaced by recognition of the number plates, a signal sent by an electronic starting gun serves as a starting point of timing, and the number plates are recognized through a computer; when the camera recognizes the athlete, processes the image, calibrates the time of each frame, records the score by taking the frame corresponding to a certain point as a judgment condition, transmits the score to a database of a computer through a USB (Universal Serial Bus), records the score, and obtains the score by reading the time on the frame corresponding to the certain point. According to the technical scheme, interference between a wearing core or a bracelet and detection equipment configured on a finishing line can be avoided, the recorded score of an athlete is more accurate, full-automatic visual identification is used for timing and circle counting, interference of human factors is reduced, and the labor cost and the system cost are reduced.

Description

Technical Field

[0001] The invention relates to the technical field of data processing, in particular to a method and system for timing and lap recording of track and field races based on computer vision. Background Art

[0002] In track and field events, traditional timing and lap recording systems are mostly electronic lap recording, and manual timing through playback, etc., and the specific use is mostly RFID radio frequency identification systems. There are detection equipment near the finish line. Athletes need to wear chips or bracelets. Radio frequency identification is used near the finish line to determine whether the athlete has crossed the line and record the number of laps. A high-speed linear array CCD camera is set at the finish line to take high-speed photos near the finish line and splice them into a long picture. A dedicated person is responsible for recording the clips of athletes crossing the line near the finish line, and the time is manually recorded through subsequent manual playback.

[0003] However, when the two athletes are close to each other and are approaching the finish line, the chips or bracelets worn by the athletes will interfere with the detection equipment at the finish line, resulting in missed laps and inaccurate final results. Manual playback timing consumes manpower and cannot achieve fully automatic timing and lap recording. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention aims to provide a method for timing and recording laps in track and field races based on computer vision, comprising:

[0005] Step 1: Acquire multiple frames of images within a preset range of the finish line of a track and field race, wherein the multiple frames of images include athletes, and splice the multiple frames of images in chronological order to obtain a finish line image;

[0006] Step 2: Determine the human skeleton structure based on the finish line image;

[0007] Step 2.1: Recognize the finish line image to determine the athlete's body contour image, and then determine the points where the human body joints are located in the body contour image;

[0008] Step 2.2: Determine the human skeleton structure of the athlete according to the points where the human joints are located;

[0009] Step 3: Recognize the human body contour image to obtain the sub-contour of the athlete's number plate and the number on the number plate;

[0010] Step 4: Mark the circles according to the sub-contour and the number on the number plate to get the number of circles corresponding to the number;

[0011] Step 5: After the number of laps corresponding to the number reaches a preset number of laps, when the number on the number plate is recognized again, the human skeleton structure is recognized to determine the shoulder points in the human skeleton structure, and then the target image corresponding to the shoulder points in the human skeleton structure is determined, and the time corresponding to the frame of the target image is obtained to obtain the score of the athlete corresponding to the number.

[0012] Optionally, step 2.2 specifically includes:

[0013] The points where the human body joints are located are spliced ​​to obtain the human body skeleton structure.

[0014] Optionally, before step 3, the following steps may be included:

[0015] The human body contour image is preprocessed, and the preprocessed image is binarized. In step 3, the binarized human body contour image is recognized.

[0016] Optionally, step 4 includes:

[0017] When a sub-contour is detected, the number of circles of the number on the number plate is increased by one to obtain the number of circles corresponding to the number.

[0018] Optionally, in step 4, a threshold time is set when circling, and after the number on the number plate is circled, the number on the number plate in the human body outline image will not be circled again within the threshold time.

[0019] Optionally, the multiple frames of images within a preset range of the track and field race finish line in step 1 are obtained by:

[0020] Capture multiple frames of images of athletes at the finish line of a track and field race;

[0021] According to a preset threshold, the collected multiple frames of images of the track and field race finish line containing athletes are intercepted to obtain multiple frames of images within a preset range of the track and field race finish line.

[0022] Optionally, all the finish line images are stored to enable review of the results.

[0023] A track and field race timing and lap-counting system based on computer vision, used for a track and field race timing and lap-counting method based on computer vision, the track and field race timing and lap-counting system based on computer vision comprising a thousand-frame camera, an electronic starting gun and a computer;

[0024] The electronic starting gun is connected to the computer, and the thousand-frame camera is connected to the computer;

[0025] The thousand-frame camera is used to collect multiple frames of images within a preset range of the track and field race finish line in real time, and transmit the multiple frames of images within the preset range of the track and field race finish line to a computer;

[0026] The electronic starting gun is used to transmit an electrical signal to a computer after the electronic starting gun is triggered;

[0027] The computer is used to receive multiple frames of images of the finish line of a track and field race in real time, and when receiving the electrical signal, uses the frame of image received at this time as the 0th frame of image; it is also used to implement the steps in the track and field race timing and lap recording method based on computer vision, and store the athletes' scores in a score management system, which is used to store the numbers and scores of all athletes.

[0028] The beneficial effects of adopting the above technical solution are:

[0029] By identifying multiple frames of images within a preset range of the finish line, mutual interference between athletes can be reduced, and the number of laps of the athlete can be determined based on the detected athlete's number. When the athlete reaches the preset number of laps, when the number on the number plate is recognized again, the human skeleton structure is identified to determine the point of the shoulder in the human skeleton structure, and then the target image corresponding to the point of the shoulder in the human skeleton structure is determined, and the time corresponding to the frame of the target image is obtained to obtain the score of the athlete corresponding to the number. By reading the time on the image, interference between the wearing core or bracelet and the detection equipment configured at the finish line can be avoided, making the recorded athlete's score more accurate, and using fully automatic visual recognition for timing and counting laps to reduce interference from human factors, and reduce labor costs and system costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 A schematic flow chart of a method for timing and recording laps in a track and field race based on computer vision in an embodiment of the present invention;

[0031] Figure 2 A schematic diagram of the connection structure of a track and field race timing and lap counting system based on computer vision in an embodiment of the present invention;

[0032] Figure 3 A schematic flow chart of another method for timing and lap recording of a track and field race based on computer vision in an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0034] In view of the problems existing in the prior art, the present invention provides a method for timing and lap recording of track and field races based on computer vision. The method uses a thousand-frame camera to identify the number plate on the outside of the athlete's thigh, and identifies the number plate to replace the current electronic lap recording. The electronic starting gun sends a signal as the starting point of timing. When the camera recognizes the athlete, it processes the image, calibrates the time of each frame, and uses the frame corresponding to a certain point as the judgment condition to record the results. The results are then transferred to the computer database via USB for entry. Figure 1 , the present invention may specifically include the following steps:

[0035] Step 1: Acquire multiple frames of images within a preset range of the finish line of a track and field race, wherein the multiple frames of images include athletes, and splice the multiple frames of images in chronological order to obtain a finish line image;

[0036] The multiple frames of images within the preset range of the track and field race finish line are obtained by:

[0037] Capture multiple frames of images of athletes at the finish line of a track and field race;

[0038] According to a preset threshold, the collected multiple frames of images of the track and field race finish line containing athletes are intercepted to obtain multiple frames of images within a preset range of the track and field race finish line.

[0039] In a specific implementation, in the multiple-frame images of the track and field race finish line containing the athlete, the finish line is vertical, and the preset threshold is the shorter distance to the left and right of the finish line. Then, when the multiple-frame images of the track and field race finish line containing the athlete are intercepted, for each frame of the image of the track and field race finish line containing the athlete, the image of the shorter distance to the left and right of the finish line is intercepted, thereby obtaining multiple-frame images within the preset range of the track and field race finish line.

[0040] By intercepting and splicing multiple frames of images of athletes at the finish line of a track and field race, the clarity of the image is improved, interference from irrelevant personnel is effectively eliminated, and the amount of computer calculation is reduced. The computer no longer needs to identify frame by frame.

[0041] Step 2: Determine the human skeleton structure based on the finish line image;

[0042] Step 2.1: Recognize the finish line image to determine the athlete's body contour image, and then determine the points where the human body joints are located in the body contour image;

[0043] In specific implementation, this step can be performed based on the trained model.

[0044] Step 2.2: Determine the human skeleton structure of the athlete according to the points where the human joints are located;

[0045] Specifically, the points where the human joints are located are spliced ​​to obtain the human skeleton structure. In a specific implementation, the present invention determines the two-dimensional positions of the points where the human joints are located, and then performs three-dimensional posture solution. According to the relationship between the points where the human joints are located, they are correctly associated and spliced ​​to obtain the human skeleton structure.

[0046] Furthermore, the human body contour image is preprocessed. Specifically, appropriate image processing is performed on the image in each contour, such as sharpening and enhancing the edge contrast, and the preprocessed image is binarized to facilitate better subsequent digit recognition.

[0047] Step 3: Recognize the human body contour image to obtain the sub-contour of the athlete's number plate and the number on the number plate;

[0048] It should be noted that the human body contour image has been processed before, so the recognition of the human body contour image in this step should be the recognition of the binarized human body contour image.

[0049] In the specific implementation, all human body contour images are traversed, and only human body contour images containing sub-contours are retained, thus eliminating the interference of irrelevant persons.

[0050] When an athlete is shown in the picture, the computer will perform image recognition, identify the human body and frame the human body range, then perform digital recognition within the framed range and frame the digital range. This not only improves the accuracy of digital recognition, but also eliminates interference from irrelevant people.

[0051] Step 4: Mark the circles according to the sub-contour and the number on the number plate to get the number of circles corresponding to the number;

[0052] Specifically, when a sub-contour is detected, the number of circles of the number on the number plate is increased by one to obtain the number of circles corresponding to the number.

[0053] When marking circles, a threshold time is set. After marking a circle on the number on the number plate, the number on the number plate in the human body contour image will not be circled again within the threshold time.

[0054] In specific implementation, the number on the number plate in the identification sub-outline is recognized. After the number on the number plate is recognized, the athlete corresponding to the number on the number plate will record one lap. Before the number on the number plate disappears completely, the lap will not be repeated. A threshold of 0.3 seconds is set. The short display and disappearance has no effect on the lap number, which can prevent multiple recognitions from causing lap number recording errors. Compared with electronic lap counting, there will be no signal interference caused by the close distance between two people, which effectively improves stability and accuracy and reduces the probability of missing records.

[0055] Step 5: After the number of laps corresponding to the number reaches a preset number of laps, when the number on the number plate is recognized again, the human skeleton structure is recognized to determine the shoulder points in the human skeleton structure, and then the target image corresponding to the shoulder points in the human skeleton structure is determined, and the time corresponding to the frame of the target image is obtained to obtain the score of the athlete corresponding to the number.

[0056] The preset number of laps is the total number of laps minus one. For example, if the total number of laps that the athlete needs to run is 8, the preset number of laps is 7.

[0057] The images in the present invention are acquired based on a 1000-frame camera, which takes 1000 pictures at a fixed rate of 1 second, so the time can be read from the image.

[0058] The present invention also stores all the finish line images to realize the review of the results. Specifically, the referee can speed up or slow down the video to conduct manual review to deal with controversial situations. After obtaining the results of all athletes or manually terminating, the results of all athletes are exported.

[0059] The present invention provides a track and field race timing and lap-counting system based on computer vision, which is used to implement a track and field race timing and lap-counting method based on computer vision. The track and field race timing and lap-counting system based on computer vision includes a thousand-frame camera, an electronic starting gun and a computer;

[0060] Combination Figure 2 , the electronic starting gun is connected to the computer, and the thousand-frame camera is connected to the computer;

[0061] The thousand-frame camera is used to collect multiple frames of images within a preset range of the track and field race finish line in real time, and transmit the multiple frames of images within the preset range of the track and field race finish line to a computer;

[0062] The electronic starting gun is used to transmit an electrical signal to a computer after the electronic starting gun is triggered;

[0063] The computer is used to receive multiple frames of images of the finish line of a track and field race in real time, and when receiving the electrical signal, uses the frame of image received at this time as the 0th frame of image; it is also used to implement the steps in the track and field race timing and lap recording method based on computer vision, and store the athletes' scores in a score management system, which is used to store the numbers and scores of all athletes.

[0064] In the specific implementation, the electronic starting gun is connected to the computer, and the 1000-frame camera is connected to the computer. When the computer and the 1000-frame camera are connected, the current screen when the software is opened is the real-time screen recorded by the camera and a vertical line running through the screen. The staff needs to adjust the position of the high-speed camera so that it faces the finish line from the side of the runway, so that the vertical line on the screen coincides with the finish line. At this time, the 1000-frame camera is in the correct position. Enter the number of competitors in the performance management system, and the preparation work is completed. Next, combine Figure 3 , the staff pulls the trigger of the electronic starting gun, emits a sound signal and transmits an electrical signal to the computer at the same time. The computer is used to receive multiple frames of images of the finish line of the track and field race in real time, and when receiving the electrical signal, the frame of image received at this time is used as the 0th frame image, and then the computer implements the steps in the track and field race timing and lap recording method based on computer vision, and stores the athletes' scores in the score management system. The score management system is used to store the numbers and scores of all athletes, and export the scores of all athletes after obtaining the scores of all athletes or manually terminating.

[0065] Among them, the present invention can also effectively solve the problem of missing laps or incorrect results due to obstruction of number plates during the recognition process by placing a camera on each side of the finish line and connecting each camera to a computer. However, since the image has been spliced, obstruction generally does not occur.

[0066] The present invention manually inputs the names of the athletes with corresponding number plates before the game, and reads the data in the database to update the performance management system and automatically sort. At the same time, it has a random grouping function for multi-person events to ensure the fairness of the event. For qualifying for the finals, the system has an automatic export function for contestants, and also classifies different events with a clear structure.

[0067] At present, machine learning technology is becoming more and more mature. In the recognition of this system, MobileNet is used for human posture estimation, and the YOLO algorithm is used for digital recognition training. Human bodies and numbers in various running events are collected as training sets to achieve the characteristics of rapid recognition and easier to correspond to the recognition in running events.

[0068] In summary, the present invention has the function of timing and counting laps for track and field races, and at the same time has the function of classifying events, entering results, and exporting promotion lists in a performance management system, thereby reducing labor costs and having a lower equipment cost than the market, making it convenient to use in small and medium-sized events.

[0069] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the embodiments of the present disclosure (but not limited to) to form a technical solution.

Claims

1. A method for timing and recording laps in track and field races based on computer vision, characterized in that: include: Step 1: Acquire multiple frames of images within a preset range of the finish line of a track and field race, wherein the multiple frames of images include athletes, and splice the multiple frames of images in chronological order to obtain a finish line image; Step 2: Determine the human skeleton structure based on the finish line image; Step 2.1: Recognize the finish line image to determine the athlete's body contour image, and then determine the points where the human body joints are located in the body contour image; Step 2.2: Determine the human skeleton structure of the athlete according to the points where the human joints are located; Step 3: Recognize the human body contour image to obtain the sub-contour of the athlete's number plate and the number on the number plate; Step 4: Mark the circles according to the sub-contour and the number on the number plate to get the number of circles corresponding to the number; Step 5: After the number of laps corresponding to the number reaches a preset number of laps, when the number on the number plate is recognized again, the human skeleton structure is recognized to determine the shoulder points in the human skeleton structure, and then the target image corresponding to the shoulder points in the human skeleton structure is determined, and the time corresponding to the frame of the target image is obtained to obtain the score of the athlete corresponding to the number.

2. The method for timing and lap recording of track and field race based on computer vision according to claim 1, characterized in that: Step 2.2 specifically includes: The points where the human body joints are located are spliced ​​to obtain the human body skeleton structure.

3. The method for timing and lap recording of track and field race based on computer vision according to claim 1, characterized in that: Also include before step 3: The human body contour image is preprocessed, and the preprocessed image is binarized. In step 3, the binarized human body contour image is recognized.

4. The method for timing and lap recording of track and field race based on computer vision according to claim 1, characterized in that: Step 4 includes: When a sub-contour is detected, the number of circles of the number on the number plate is increased by one to obtain the number of circles corresponding to the number.

5. The method for timing and lap recording of track and field race based on computer vision according to claim 1, characterized in that: In step 4, when marking circles, a threshold time is set. After marking a circle on the number on the number plate, the number on the number plate in the human body contour image will not be circled again within the threshold time.

6. The method for timing and lap recording of track and field race based on computer vision according to claim 1, characterized in that: The multiple frames of images within the preset range of the track and field race finish line in step 1 are obtained by: Capture multiple frames of images of athletes at the finish line of a track and field race; According to a preset threshold, the collected multiple frames of images of the track and field race finish line containing athletes are intercepted to obtain multiple frames of images within a preset range of the track and field race finish line.

7. The method for timing and lap recording of track and field race based on computer vision according to claim 1, characterized in that: All the finish line images are stored to enable review of the results.

8. A track and field race timing and lap recording system based on computer vision, used to implement the track and field race timing and lap recording method based on computer vision according to claim 1, characterized in that: The track and field race timing and lap recording system based on computer vision comprises a thousand-frame camera, an electronic starting gun and a computer; The electronic starting gun is connected to the computer, and the thousand-frame camera is connected to the computer; The thousand-frame camera is used to collect multiple frames of images within a preset range of the track and field race finish line in real time, and transmit the multiple frames of images within the preset range of the track and field race finish line to a computer; The electronic starting gun is used to transmit an electrical signal to a computer after the electronic starting gun is triggered; The computer is used to receive multiple frames of images of the finish line of a track and field race in real time, and when receiving the electrical signal, uses the frame of image received at this time as the 0th frame of image; it is also used to implement the steps in the track and field race timing and lap recording method based on computer vision, and store the athletes' scores in a score management system, which is used to store the numbers and scores of all athletes.

Images