Track race evaluation method
By collecting multi-frame images to identify the positional relationship between the target individual and the starting line and the finish line, combined with image tracking technology, the problem of participants' lane-crossing and relay cheating in track competitions is solved, and automated and accurate performance evaluation is achieved.
Patent Information
- Application Number
- CN202311852804.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-29
- Publication Date
- 2025-07-01
AI Technical Summary
In the existing track and track competition evaluation methods, there are problems of contestants cheating in the track and relay, resulting in inaccurate results, and the existing machine vision-based evaluation solutions are difficult to effectively solve.
By collecting multi-frame images, identifying the target individual and establishing their positional relationship with the starting line and finishing line, combining image tracking technology, obtaining accurate starting and arrival time, determining whether the runway number is correct, and achieving automated evaluation.
It improves the accuracy and persuasiveness of the evaluation results, reduces manual participation, and is suitable for human-organized events and unorganized self-training.
Abstract
Description
Technical Field
[0001] The present invention relates to a track event evaluation method. Background Art
[0002] For example, the performance evaluation of track events such as 50-meter and 100-meter races is becoming increasingly intelligent. The test method based on machine vision is one of them, which has many advantages. It does not require wearing devices for assistance, nor does it need to bury sensors underground, so the layout cost is relatively low. However, in the prior art, the performance is mainly evaluated based on the following method: obtaining the list of participants and the starting time on the starting side, then associating the target individuals in the list of participants with the track, then obtaining the arrival time of the person on the corresponding track on the ending side, and then giving the result of the corresponding track to the target individual associated with the track. This scheme is relatively simple, but there are certain problems. For example, if someone cuts in or cheats in the relay, the result will be incorrect and needs to be improved urgently. Summary of the Invention
[0003] To solve the technical problems in the background art, the present invention provides a track event evaluation method, including:
[0004] Collecting multiple frames of first images, identifying target individuals from the first images, and identifying first key frames from the first images. There is a preset positional relationship between the first moving individuals and the first identifiers in the first key frames, where the first moving individuals are identified as target individuals through tracking.
[0005] "The first moving individuals are identified as target individuals through tracking" specifically includes:
[0006] After identifying the first moving individuals in the first key frames as target individuals, tracking the target individuals until the first key frames are identified.
[0007] "The first moving individuals are identified as target individuals through tracking" specifically includes:
[0008] Tracking the moving individuals appearing in the first images, obtaining the key frames of the moving individuals, and there is a preset positional relationship between the first identifiers and the moving individuals in the key frames;
[0009] Continuously identifying the first images, identifying the first moving individuals that are the target individuals from the moving individuals, and the key frames corresponding to the first moving individuals are the first key frames.
[0010] The target individuals are in the list of participants, and the reference list is actively input or system-assigned.
[0011] It further includes:
[0012] Obtaining the starting time of the target individuals; obtaining the arrival time of the target individuals; evaluating the performance of the target individuals according to the starting time and the arrival time.
[0013] The starting time is associated with the starting signal time. The first image is an image on the finish side. "Obtaining the arrival time of the target individual" specifically means: analyzing a second key frame in which the target individual has a preset relationship with the finish line from the first image, and associating the arrival time based on the acquisition time of the second key frame.
[0014] It further includes: collecting multiple frames of second images, where the second images are images on the starting side; "obtaining the starting time of the target individual" specifically means: analyzing a third key frame in which the target individual has a preset relationship with the starting line from the second images, and associating the starting time based on the acquisition time of the third key frame; "obtaining the arrival time of the target individual" specifically means: analyzing a second key frame in which the target individual has a preset relationship with the finish line from the first image, and associating the arrival time based on the acquisition time of the second key frame.
[0015] It further includes:
[0016] Obtaining the first lane number where the target individual is located when starting; obtaining the second lane number where the target individual is located when arriving, and comparing the first lane number with the second lane number to determine whether the target individual has switched lanes.
[0017] The first lane number is obtained by active input or system allocation or from the analysis of the third key frame; the second lane number is obtained through the analysis of the first image.
[0018] It further includes:
[0019] Tracking the target individual in multiple frames of the first image to determine whether the target individual has switched lanes on the finish side.
[0020] The motion evaluation method proposed by the present invention has the following technical effects: when evaluating track and field events, both the starting time and the arrival time are strongly associated with the target individual, and there will be no misattribution due to lane switching. Compared with the solution of binding the target individual to the lane number, obviously the evaluation result is more persuasive and accurate. At the same time, the proposed solution also makes the evaluation process less dependent on manual participation. Therefore, whether it is for organized events or self-training without organization, the effect will be better. Detailed implementation mode
[0021] The present invention proposes a track and field event evaluation method, including:
[0022] S1. Obtaining the identity information, starting time, and the first lane number where the target individual is located when starting as a participant in the race.
[0023] Specifically included:
[0024] S11. Collecting multiple frames of second images, where the second images are images on the starting side, and identifying the target individual participating in the race from the second images.
[0025] Of course, as an alternative solution, the identity information of the target individuals participating in the race can also be obtained by actively inputting or system - allocating the list of participants in the race. This method belongs to the category of existing technologies and will not be elaborated here.
[0026] S12. Analyze the third key frame in which the target individual has a preset relationship with the starting line from the second image, associate the starting time based on the acquisition time of the third key frame, and obtain the first lane number from the analysis of the third key frame.
[0027] Among them, "the target individual has a preset relationship with the starting line" specifically means that the distance between the center point of the bottom edge of the target individual's body frame and the starting line is less than a preset threshold, which can be understood as the target individual reaching or exceeding the starting line position.
[0028] As an alternative solution: the starting time of the target individual is associated with the starting command time, and the first lane is actively input or system - allocated.
[0029] S2. Collect multiple frames of the first image, identify the target individual from the first image, and identify the first key frame from the first image. There is a preset positional relationship between the first moving individual and the first identifier in the first key frame. Among them, the first moving individual is recognized as the target individual through tracking, and the first image is an image on the finish line side.
[0030] There are two implementation methods for "the first moving individual is recognized as the target individual through tracking":
[0031] Method 1: After recognizing the first moving individual in the first key frame as the target individual, track the target individual until the first key frame is recognized.
[0032] Method 2: Track the moving individuals appearing in the first image, obtain the key frames of the moving individuals, and there is a preset positional relationship between the first identifier and the moving individual in the key frames; continuously identify the first image, and identify the first moving individual, which is the target individual, from the moving individuals. The key frame corresponding to the first moving individual is the first key frame.
[0033] To save costs and facilitate layout, in the present invention, on the finish line side, only one acquisition terminal on the finish line side is used to collect the first image. It is necessary to collect the first image that contains the facial features of the target individual and is clear enough, and it is also necessary to collect the first key frame. To collect the first key frame, the image acquisition area of the acquisition terminal on the finish line side needs to cover at least the finish line.
[0034] Since in the second key frame, it is very difficult to ensure that the target individual always faces the acquisition terminal on the finish line side. If forced, it will not only affect the user experience, but may also cause the target individual to perform abnormally due to distraction, which is difficult to solve. The solution proposed by the present invention does not require identifying the target individual in the first key frame. Instead, the field of view of the acquisition terminal on the finish line side is expanded to at least cover the finish line and at least 3.5 meters of the runway in front of the finish line. The field of view of the acquisition terminal on the finish line side covering the finish line can meet the acquisition of the first key frame. The present invention does not need to identify the target individual from the first key frame because, as previously described, it is very difficult to ensure that the frame image must contain face features with qualified angles and clarity, and in actual situations, in the vast majority of second key frames, there are no face features with qualified angles and clarity.
[0035] The solution proposed by the present invention lies in:
[0036] The acquisition terminal on the finish line side acquires multiple first images. Only one frame of the multiple first images with qualified face feature angles and clarity is required. Combining with tracking, this frame of the first image that can identify the target individual can be associated with the first key frame corresponding to the target individual.
[0037] S3. Obtain the arrival time of the target individual and the second runway number where the target individual is located when arriving.
[0038] S31. Analyze the second key frame in which the target individual has a preset relationship with the finish line from the first images, and associate the arrival time according to the acquisition time of the second key frame.
[0039] "The target individual has a preset relationship with the finish line" specifically means that the distance between the center point of the bottom edge of the target individual's body frame and the starting line is less than a preset threshold, which can be understood as the target individual reaching or exceeding the starting line position.
[0040] S32. Obtain the second runway number by analyzing the first images or the second key frame.
[0041] S4. Compare the first runway number with the second runway number to determine whether the target individual has switched lanes.
[0042] S5. Evaluate the performance of the target individual according to the starting time and the arrival time.
[0043] S6. Track the target individual in multiple first images to determine whether the target individual has switched lanes on the finish line side.
[0044] In addition, for the entire solution, both the starting time and the arrival time are strongly associated with the target individual, and there will be no misattribution due to lane-changing. Compared with the solution where the target individual is bound to the lane number, the evaluation results are obviously more persuasive and accurate. At the same time, the proposed solution also requires less manual intervention in the evaluation process, making it more effective whether it is for organized competitions or self-training without organization.
[0045] The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent substitutions or changes, should be covered by the protection scope of the present invention.
Claims
1. A track and field event evaluation method, characterized in that, Including: Collect multiple first images, identify a target individual from the first images, and identify a first key frame from the first images. There is a preset positional relationship between a first moving individual and a first identifier in the first key frame, where the first moving individual is identified as the target individual through tracking.
2. The evaluation method according to claim 1, wherein "The first moving individual is identified as the target individual through tracking" specifically includes: After identifying the first moving individual in the first key frame as the target individual, track the target individual until the first key frame is identified.
3. The evaluation method according to claim 1, characterized in that "The first moving individual is identified as the target individual through tracking" specifically includes: Track the moving individuals that appear in the first image, obtain the key frames of the moving individuals, and there is a preset positional relationship between the first identifier and the moving individuals in the key frames; Continuously identify the first image, and identify the first moving individual that is the target individual from the moving individuals. The key frame corresponding to the first moving individual is the first key frame.
4. The evaluation method according to claim 1, wherein The target individual is on the participating list, and the reference list is actively entered or system-assigned.
5. The evaluation method according to claim 1, characterized in that Also including: Obtain the starting time of the target individual; obtain the arrival time of the target individual; evaluate the performance of the target individual based on the starting time and the arrival time.
6. The evaluation method according to claim 5, wherein The starting time is associated with the starting command time. The first image is an image on the finish side. "Obtain the arrival time of the target individual" specifically means: Analyze a second key frame in which the target individual has a preset relationship with the finish line from the first image, and associate the arrival time based on the acquisition time of the second key frame.
7. The evaluation method according to claim 5, wherein Also including: Collect multiple second images, and the second images are images on the starting side; "Obtain the starting time of the target individual" specifically means: Analyze a third key frame in which the target individual has a preset relationship with the starting line from the second image, and associate the starting time based on the acquisition time of the third key frame; "Obtain the arrival time of the target individual" specifically means: Analyze a second key frame in which the target individual has a preset relationship with the finish line from the first image, and associate the arrival time based on the acquisition time of the second key frame.
8. The evaluation method according to claim 7, wherein Also including: Obtain the first lane number where the target individual is located at the start; Obtain the second lane number where the target individual is located at the arrival, and compare the first lane number with the second lane number to determine whether the target individual has changed lanes.
9. The evaluation method according to claim 8, wherein The first lane number is actively entered or system-assigned or obtained from the analysis of the third key frame; the second lane number is obtained through the analysis of the first image.
10. The evaluation method according to claim 9, wherein Also including: Track the target individual in multiple first images to determine whether the target individual has changed lanes on the finish side.