Sprint score test method and system based on image segmentation

Abstract

The invention discloses a sprint score testing method and system based on image segmentation, and relates to the technical field of sprint score testing. The method comprises the following steps of: firstly, importing original image data captured by a plurality of players in real time by image acquisition equipment into a pre-trained multi-target image segmentation model in real time to obtain a real-time segmentation result of a plurality of body parts of each player; a corresponding trunk pixel point set is extracted in real time from the real-time segmentation result of the corresponding trunk part, and if it is found for the first time that a non-empty intersection exists between a certain corresponding trunk pixel point set and the end point end face set of the corresponding sprint track, the corresponding trunk pixel point set is extracted; and if so, recording a first image acquisition moment corresponding to the certain trunk pixel point set as a corresponding actual line punching moment, and calculating a time difference between the actual line punching moment and the actual start moment to obtain a corresponding actual sprint score, so that the measurement precision of the sprint test score can be improved to the maximum extent, and errors can be reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of sprint testing, and in particular relates to a sprint performance testing method and system based on image segmentation. Background Art

[0002] As society places increasing emphasis on student health management, scientific methods for assessing student physical fitness have become a key focus for researchers in related fields. As a key indicator of student physical fitness, the 50-meter run fully demonstrates a student's lower limb explosive power, reaction speed, agility, coordination, and muscle endurance. Therefore, a scientific 50-meter run test is essential. Currently, traditional sprint performance testing methods for the 50-meter run rely entirely on manual labor, requiring a starting flag and stopwatch to complete the test at two locations (i.e., the starting and finish points). Because the 50-meter running speeds of different test subjects vary slightly, only two people can be tested at a time. Consequently, traditional sprint performance testing methods suffer from low accuracy, large manual timing errors, long testing time, high workload for testers, and difficulty in identifying violations. Consequently, the resulting performance data cannot accurately reflect students' physical fitness.

[0003] Existing patent 201621307970.7 provides a running detection system based on depth algorithm image acquisition, including a bracket, a control host and at least one depth algorithm image acquisition device. The depth algorithm image acquisition device is fixedly mounted on the bracket and the camera is opposite to the end of the runway. The depth algorithm image acquisition device is connected to the control host via a wired or wireless connection. When the athlete starts running, the control host starts timing. When an athlete reaches the finish line, the depth algorithm image acquisition device will detect that there is an object within the runway, 3 meters away from the camera, and the timing of this runway will stop. In this way, the time difference is recorded as the athlete's score. However, this testing method conflicts with the actual method of determining the finish line, and is prone to test errors and no-foul detection problems.

[0004] Existing patent 201210402201.5 provides a multi-purpose running imaging timing system, which relates to the field of track and field timing technology. It includes a starting section and a finishing section. The starting section consists of an infrared foul detection device and a wireless starter. The finishing section includes a wireless receiver, a high-speed integrated dome camera, a finish control unit, and a timing computer terminal. When used in multi-lap middle- and long-distance runs, the finish section also includes an infrared sprint detection device and an RFID (Radio Frequency Identification) card reader. Runners wear RFID-tagged vests with numbers. The infrared cross-line detection device emits infrared signal beams that coincide with the finish line to detect the runner's sprint moment, triggering the finish control unit and the timing computer terminal to synchronously record and lock the runner's score. Each time a runner crosses the finish line, the card reader reads the RFID tag information and transmits it to the finish control unit and the timing computer terminal to automatically calculate the lap number. However, this testing method still has errors, requires a lot of equipment, is costly, and lacks foul detection.

[0005] Based on the aforementioned patented technologies, it can be seen that current sprint testing methods primarily rely on manual, electrical, and intelligent testing. In recent years, with the rise of artificial intelligence (AI), patented technologies that utilize these technologies to test running performance have become a trend. However, these methods still have some errors in performance measurement and are relatively crude in detecting fouls. For example, the widely adopted key-point-based sprint performance testing method suffers from significant errors, ignores the human body's contours, and reduces measurement accuracy. Furthermore, according to IAAF regulations, athletes crossing the finish line in track and field sprints should be considered based on the torso area contacting the finish line. In other words, as long as the first contact is with the abdomen, chest, or back, the result can be determined. Therefore, existing key-point-based sprint performance testing methods struggle to accurately determine the specific location where a competitor crossed the finish line. Summary of the Invention

[0006] The purpose of the present invention is to provide a sprint performance test method, system, computer device, computer-readable storage medium and computer program product based on image segmentation, so as to solve the problems of low measurement accuracy and large errors caused by ignoring the human body contour in existing sprint performance test schemes.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] In a first aspect, a sprint performance testing method based on image segmentation is provided, which is performed by a computer device that is communicatively connected to a competition notification device and an image acquisition device, and includes:

[0009] Obtaining a one-to-one correspondence between a plurality of sprint tracks and a plurality of contestants, wherein the plurality of sprint tracks are sequentially adjacent and constitute a sprint stadium;

[0010] After the multiple contestants enter the pre-starting position of the sprint stadium, raw image data captured by the image acquisition device for the multiple contestants are received in real time, and the raw image data are imported into a pre-trained multi-objective image segmentation model in real time to obtain real-time segmentation results of multiple body parts of each of the multiple contestants, wherein the image acquisition device includes a depth camera for obtaining depth information and / or a three-dimensional camera for capturing high-speed and high-resolution images of the contestants in real time, and the multiple body parts include a head, a torso, and a limb.

[0011] For each of the contestants, a corresponding torso pixel point set is extracted in real time from the real-time segmentation results of the corresponding torso, wherein the coordinates of the pixel points in the torso pixel point set are three-dimensional coordinates in a three-dimensional coordinate system of the competition field, wherein the X-axis direction of the three-dimensional competition field coordinate system is perpendicular to the sprint track, the Y-axis direction of the three-dimensional competition field coordinate system is the direction from the starting point to the finish line, and the Z-axis direction of the three-dimensional competition field coordinate system is perpendicular to the boundary line of the sprint track and passes through the starting point;

[0012] When the competition notification device issues a start signal, recording the time when the start signal is issued as the actual start time;

[0013] After the competition notification device sends the starting signal, for each competitor, if a non-empty intersection is first found between a corresponding torso pixel point set and a corresponding end face set of the sprint track in the three-dimensional coordinate system of the competition field, the first image acquisition time corresponding to the torso pixel point set is recorded as the corresponding actual finishing time T ecm , and calculate the actual finishing time T ecm The actual starting time T esm The time difference is used to obtain the corresponding actual sprint results.

[0014] Based on the above invention content, a new solution for completing a sprint performance test based on a multi-target image segmentation model is provided, namely, the original image data captured in real time by the image acquisition device for multiple contestants are first imported into the pre-trained multi-target image segmentation model in real time to obtain real-time segmentation results of multiple body parts of each contestant, and then for each contestant, the corresponding torso pixel point set is extracted in real time from the real-time segmentation results of the corresponding torso, and if a corresponding torso pixel point set is found to have a non-empty intersection with the corresponding end face set of the sprint track for the first time, the first image acquisition moment corresponding to the torso pixel point set is recorded as the corresponding actual finishing moment, and the time difference between the actual finishing moment and the actual starting moment is calculated to obtain the corresponding actual sprint performance. In this way, the measurement accuracy of the sprint test performance can be maximized and the error can be reduced, which is convenient for practical application and promotion.

[0015] In one possible design, the image acquisition device includes three depth cameras for acquiring depth information and two three-dimensional cameras for capturing high-speed, high-resolution images of the contestants in real time, wherein the first depth camera is arranged on the upper end face of the starting range of the sprint stadium with its lens facing the sprint stadium, the second depth camera is arranged above the main running range of the sprint stadium with its lens facing the sprint stadium, the third depth camera is arranged on the lower end face of the finishing range of the sprint stadium with its lens facing the sprint stadium, the first three-dimensional camera is arranged on the lower end face of the starting range with its lens facing the sprint stadium, the second three-dimensional camera is arranged on the upper end face of the finishing range with its lens facing the sprint stadium, and the main running range is located between the starting range and the finishing range.

[0016] In one possible design, after receiving in real time the raw image data captured by the image acquisition device for each of the multiple contestants in real time, the method further includes:

[0017] For any one of the plurality of contestants, recording an image sequence of the corresponding contestant within the starting range and / or the finishing range, wherein the image sequence comprises a plurality of image groups that are sequentially sequential in image acquisition time sequence, and the image groups comprise a plurality of image frames acquired by different plurality of image acquisition devices at the same image acquisition time;

[0018] For each frame of the multiple images, calculate the location of the contestant at the corresponding image capture moment based on a corresponding set of full-body pixel points extracted from the real-time segmentation results of multiple body parts of the contestant, and further fuse the depth data of the corresponding image capture device to generate a three-dimensional point cloud of the contestant at the corresponding image capture moment from the perspective of the corresponding image capture device;

[0019] For each image group in the image sequence, using multi-view stereo vision technology to fuse the three-dimensional point clouds of the contestant at the corresponding image capture moment from different perspectives, and generate a three-dimensional model of the contestant at the corresponding image capture moment through triangulation;

[0020] The position and three-dimensional model of any contestant at each image acquisition moment are summarized to obtain a three-dimensional model reconstruction result of any contestant in the starting range and / or the finishing range.

[0021] In one possible design, after obtaining the three-dimensional model reconstruction result of any one of the contestants within the starting range and / or the finishing range, the method further includes:

[0022] The 3D visualization tool OpenGL is used to display the three-dimensional model reconstruction result of any contestant in the starting range and / or the finishing range, so as to conduct a three-dimensional starting / finishing review of any contestant after the race.

[0023] In one possible design, the multi-object image segmentation model is pre-trained based on the Mask R-CNN model.

[0024] In one possible design, after obtaining the real-time segmentation results of the multiple body parts of each contestant, the method further includes:

[0025] For each contestant, extracting in real time a corresponding whole-body pixel point set from the real-time segmentation results of the corresponding multiple body parts, wherein the coordinates of the pixel points in the whole-body pixel point set are three-dimensional coordinates in the three-dimensional coordinate system of the competition field;

[0026] After the competition notification device sends a start signal, for each contestant, if a corresponding whole-body pixel point set is first found to have a non-empty intersection with the starting end face set of the corresponding sprint track in the three-dimensional coordinate system of the competition field, the second image acquisition time corresponding to the whole-body pixel point set is recorded as the corresponding sensory start time T fsm , and if the starting time of the sense is determined to be T fsm The actual starting time Tesm If the time difference is less than the preset time, the corresponding athlete is judged to have committed a false start violation.

[0027] In one possible design, after obtaining the real-time segmentation results of the multiple body parts of each contestant, the method further includes:

[0028] For each contestant, extracting in real time a corresponding whole-body pixel point set from the real-time segmentation results of the corresponding multiple body parts, wherein the coordinates of the pixel points in the whole-body pixel point set are three-dimensional coordinates in the three-dimensional coordinate system of the competition field;

[0029] After the competition notification device sends a starting signal, for each contestant, if a corresponding set of whole-body pixel points is found for the first time to have a non-empty intersection with the boundary surface set of the corresponding sprint track in the three-dimensional coordinate system of the stadium, it is determined that the corresponding contestant has committed a lane-crossing violation.

[0030] In one possible design, after obtaining the real-time segmentation results of the multiple body parts of each contestant, the method further includes:

[0031] For each contestant, extracting in real time a corresponding whole-body pixel point set from the real-time segmentation results of the corresponding multiple body parts, wherein the coordinates of the pixel points in the whole-body pixel point set are three-dimensional coordinates in the three-dimensional coordinate system of the competition field;

[0032] After the competition notification device sends a starting signal, for each contestant, if a corresponding set of whole-body pixel points is found for the first time to have a non-empty intersection with the lower limit end face set of the finishing range of the corresponding sprint track in the three-dimensional coordinate system of the stadium, the competition notification device is triggered to perform a finishing line warning reminder action.

[0033] In a second aspect, a sprint performance testing system based on image segmentation is provided, which is suitable for being arranged in a computer device that is communicatively connected to a competition notification device and an image acquisition device, and includes a correspondence acquisition unit, an image segmentation processing unit, a torso pixel extraction unit, a start time recording unit, and a sprint performance measurement unit.

[0034] The correspondence obtaining unit is configured to obtain a one-to-one correspondence between a plurality of sprint tracks and a plurality of contestants, wherein the plurality of sprint tracks are adjacent to each other and constitute a sprint stadium;

[0035] The image segmentation processing unit is configured to receive, in real time, raw image data captured by the image acquisition device of the multiple contestants in real time after the multiple contestants enter the pre-starting position of the sprint stadium, and import the raw image data into a pre-trained multi-objective image segmentation model in real time to obtain real-time segmentation results of multiple body parts of each of the multiple contestants, wherein the image acquisition device includes a depth camera for obtaining depth information and / or a three-dimensional camera for capturing high-speed and high-resolution images of the contestants in real time, and the multiple body parts include a head, a torso, and a limb.

[0036] The torso pixel point extraction unit is communicatively connected to the image segmentation processing unit and is configured to extract, in real time, a corresponding torso pixel point set from the real-time segmentation results of the corresponding torso for each contestant, wherein the coordinates of the pixel points in the torso pixel point set are three-dimensional coordinates in a three-dimensional coordinate system of the competition field, wherein the X-axis direction of the three-dimensional coordinate system of the competition field is perpendicular to the sprint track, the Y-axis direction of the three-dimensional coordinate system of the competition field is the direction from the starting point to the finish line, and the Z-axis direction of the three-dimensional coordinate system of the competition field is perpendicular to the boundary line of the sprint track and passes through the starting point;

[0037] The starting time recording unit is configured to record the time when the competition notification device sends the starting signal as the actual starting time;

[0038] The sprint performance measurement unit is communicatively connected to the corresponding relationship acquisition unit, the torso pixel point extraction unit, and the starting time recording unit, and is configured to, after the competition notification device issues a starting signal, for each competitor, if a non-empty intersection is first found between a corresponding torso pixel point set and a corresponding end face set of the sprint track in the three-dimensional coordinate system of the competition field, record the first image acquisition time corresponding to the torso pixel point set as the corresponding actual finishing time T ecm , and calculate the actual finishing time T ecm The actual starting time T esm The time difference is used to obtain the corresponding actual sprint results.

[0039] In a third aspect, a sprint performance testing system based on image segmentation is provided, comprising a competition notification device, an image acquisition device, and a computer device;

[0040] The race notification device is communicatively connected to the computer device, and is used to send a start signal and trigger the computer device to record the actual start time;

[0041] The image acquisition device is communicatively connected to the computer device and is used to capture raw image data of multiple contestants in real time and transmit the captured results to the computer device in real time;

[0042] The computer device is used to execute the sprint performance testing method as described in the first aspect or any possible design of the first aspect.

[0043] In a fourth aspect, the present invention provides a computer device comprising a memory, a processor, and a transceiver communicatively connected in sequence, wherein the memory is used to store a computer program, the transceiver is used to send and receive messages, and the processor is used to read the computer program and execute the sprint performance testing method as described in the first aspect or any possible design of the first aspect.

[0044] In a fifth aspect, the present invention provides a computer-readable storage medium having instructions stored thereon. When the instructions are run on a computer, the sprint performance testing method as described in the first aspect or any possible design of the first aspect is executed.

[0045] In a sixth aspect, the present invention provides a computer program product, comprising a computer program or instructions, which, when executed by a computer, implements the sprint performance testing method as described in the first aspect or any possible design of the first aspect.

[0046] Beneficial effects of the above scheme:

[0047] (1) The present invention provides a new solution for completing a sprint performance test based on a multi-target image segmentation model, namely, firstly, the original image data captured by the image acquisition device for a plurality of contestants in real time are imported into a pre-trained multi-target image segmentation model in real time to obtain real-time segmentation results of a plurality of body parts of each contestant, and then, for each contestant, the corresponding torso pixel point set is extracted in real time from the real-time segmentation results of the corresponding torso, and if a non-empty intersection is found for the first time between a corresponding torso pixel point set and a corresponding end face set of the sprint track, the first image acquisition moment corresponding to the certain torso pixel point set is recorded as the corresponding actual finishing moment, and the time difference between the actual finishing moment and the actual starting moment is calculated to obtain the corresponding actual sprint performance, so as to maximize the measurement accuracy of the sprint test performance and reduce the error;

[0048] (2) It can also monitor illegal behaviors such as false starts and lane-crossing, and provide early warnings for finish line violations, ensuring maximum monitoring accuracy;

[0049] (3) It is also possible to reconstruct a three-dimensional model of the contestant within the starting range and / or the finishing range to facilitate a three-dimensional start / finish review after the race, which helps the coaching system train the contestants’ weaknesses and further improves their practicality;

[0050] (4) It can also obtain more accurate and fairer competition results, and help contestants review and analyze the results. It can be used not only for student physical tests, but also for competitive tests and track and field events, facilitating practical application and promotion. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 A flowchart of a sprint performance testing method based on image segmentation provided in an embodiment of the present application.

[0053] Figure 2 This is an example diagram of the positional relationship between a sprint stadium and multiple image acquisition devices provided in an embodiment of the present application.

[0054] Figure 3 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the present invention will be briefly introduced below in conjunction with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structures of the drawings is only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained based on these embodiments without creative work. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.

[0056] It should be understood that although the terms first, second, etc. may be used herein to describe various objects, these objects should not be limited by these terms. These terms are merely used to distinguish one object from another. For example, a first object can be referred to as a second object, and similarly, a second object can be referred to as a first object without departing from the scope of the exemplary embodiments of the present invention.

[0057] It should be understood that the term "and / or" that may appear in this document is merely a description of the association relationship between associated objects, indicating that there may be three relationships. For example, A and / or B can indicate three situations: A exists alone, B exists alone, or A and B exist at the same time. For another example, A, B and / or C can indicate the existence of any one of A, B and C or any combination of them. The term " / and" that may appear in this document describes another type of association object relationship, indicating that there may be two relationships. For example, A / and B can indicate two situations: A exists alone or A and B exist at the same time. In addition, the character " / " that may appear in this document generally indicates that the previous and next associated objects are in an "or" relationship.

[0058] Example

[0059] like Figure 1 As shown, the sprint performance test method based on image segmentation provided in the first aspect of this embodiment can be, but is not limited to, executed by a computer device having certain computing resources and being respectively connected to the competition notification device and the image acquisition device, such as a cloud server, an edge computer equipped with a GPU, a personal computer (PC, a multi-purpose computer with a size, price and performance suitable for personal use; desktops, laptops, small laptops, tablets and ultrabooks are all personal computers), a smart phone, a personal digital assistant (PDA) or a wearable device. Figure 1 As shown, the sprint performance test method may include, but is not limited to, the following steps S1 to S5.

[0060] S1. Obtain a one-to-one correspondence between a plurality of sprint tracks and a plurality of contestants, wherein the plurality of sprint tracks are adjacent to each other and constitute a sprint competition venue.

[0061] In step S1, the sprint track is a track provided for contestants (e.g., students to be tested for physical fitness) to compete in sprint events such as 50-meter run, 60-meter run, or 100-meter run; Figure 2As shown, the sprint stadium is composed of K (K represents a positive integer greater than or equal to 3) adjacent sprint tracks. The specific method of obtaining the multiple sprint tracks and the multiple contestants may include, but is not limited to: before the start of the game, extracting the contestants' number plate information, and collecting the attribute data of each contestant to obtain a registration information set R and register it, wherein the specific format of the registration information set R can be {(name, number plate), ..}; then processing the registration information set R to convert it into a structure that is convenient for subsequent processing, and obtaining a number plate list N = {N1, N2, ..., N} that reflects the one-to-one correspondence between the multiple sprint tracks and the multiple contestants. k ,…,N K}, where k is a positive integer less than or equal to K, and N k Indicates the number plate of the kth competitor corresponding to the kth sprint track.

[0062] S2. After the multiple contestants enter the pre-starting positions of the sprint stadium, the original image data captured by the image acquisition device for the multiple contestants are received in real time, and the original image data are imported into a pre-trained multi-target image segmentation model in real time to obtain real-time segmentation results of multiple body parts of each of the multiple contestants, wherein the image acquisition device includes but is not limited to a depth camera for obtaining depth information and / or a three-dimensional camera for capturing high-speed and high-resolution images of the contestants in real time, and the multiple body parts include but are not limited to the head, torso and limbs.

[0063] In step S2, the preparatory starting position of the sprint track is as follows: Figure 2 The image acquisition device is used to capture the original image data of the multiple contestants in real time based on the existing image acquisition technology, and transmit the captured results to the local device in real time. Figure 2 As shown, the image acquisition device includes but is not limited to three depth cameras for acquiring depth information and two three-dimensional cameras for capturing high-speed and high-resolution images of contestants in real time, wherein the first depth camera (i.e. Figure 2 The first camera in the figure) is arranged on the upper end face of the starting range of the sprint stadium and faces the sprint stadium, and the second depth camera (i.e. Figure 2 The 2nd camera in the figure) is arranged above the main running range of the sprint stadium and the lens is directed toward the sprint stadium, and the third depth camera (i.e. Figure 2 3) is arranged on the lower end face of the sprint finish range and faces the sprint field. The first three-dimensional camera (i.e. Figure 2The 4th camera in the figure) is arranged on the lower end face of the starting range and faces the sprint field, and the second 3D camera (i.e. Figure 2 The camera No. 5 in the figure is arranged on the upper end face of the finish line range and the lens is directed toward the sprint field. The main running range is located between the starting range and the finish line range. Figure 2 As shown, the starting range, the main running range, and the finishing range constitute the entire sprint distance. For example, if the sprint event is a 50-meter race, the starting range can be the area corresponding to the [0, 3]-meter distance, the main running range can be the area corresponding to the (3, 47)-meter distance, and the finishing range can be the area corresponding to the [47, 50]-meter distance. To prevent cameras 1-5 from adversely affecting the competition, they can be placed at a certain height above the ground, for example, 5 meters above the ground.

[0064] In step S2, the original image data will have multiple copies in time sequence and correspond to different image acquisition moments, so the final image capture results will constitute an image sequence, and the real-time segmentation results of the multiple body parts of each contestant will also constitute a time series data. The multi-target image segmentation model is used to perform image segmentation processing on the corresponding part of the input image data for each body part among the multiple body parts. It can be obtained by pre-training based on, but not limited to, the Mask R-CNN model, wherein the Mask R-CNN (Mask Region-based Convolutional Neural Network) model is an existing deep learning model for target detection and instance segmentation. It is an extended version of the Faster R-CNN model and can generate a binary mask of the target to achieve accurate instance segmentation. Therefore, the multi-target image segmentation model can be obtained by training based on certain image sample data through conventional pre-training methods. In addition, the limb part can also be subdivided into left leg, right leg, left arm and right arm, etc.

[0065] S3. For each of the contestants, a corresponding torso pixel point set is extracted in real time from the real-time segmentation results of the corresponding torso, wherein the coordinates of the pixel points in the torso pixel point set are three-dimensional coordinates in the three-dimensional coordinate system of the stadium, the X-axis direction of the three-dimensional coordinate system of the stadium is perpendicular to the sprint track, the Y-axis direction of the three-dimensional coordinate system of the stadium is the direction from the starting point to the finish line, and the Z-axis direction of the three-dimensional coordinate system of the stadium is perpendicular to the boundary line of the sprint track and passes through the starting point.

[0066] In step S3, since the image acquisition device includes a depth camera for acquiring depth information and / or a 3D camera for capturing high-speed and high-resolution images of the contestants in real time, the original image data and the real-time segmentation results of the multiple body parts will contain depth information and / or 3D coordinate information, and the torso pixel point set can be conventionally extracted from the real-time segmentation results of the torso. Figure 2 As shown, at this time, if the unit length of the three-dimensional coordinate system of the stadium is 1 meter, the lower limit end face of the starting range can be expressed as (x, -3, z), the upper limit end face of the starting range can be expressed as (x, 0, z), the lower limit end face of the finish line range can be expressed as (x, 44, z), and the upper limit end face of the finish line range can be expressed as (x, 47, z).

[0067] S4. When the game notification device issues a start signal, the time when the start signal is issued is recorded as the actual start time T esm .

[0068] In step S4, the competition notification device is used to send out a start signal, and because it is connected to the local device through communication, it can conventionally trigger the local device to record the time of sending out the start signal as the actual start time T when the competition notification device sends out the start signal. esm .

[0069] S5. After the competition notification device issues the starting signal, for each competitor, if a corresponding torso pixel point set is first found to have a non-empty intersection with the corresponding sprint track finish end face set in the three-dimensional coordinate system of the competition field, the first image acquisition time corresponding to the torso pixel point set is recorded as the corresponding actual finishing time T. ecm , and calculate the actual finishing time T ecm The actual starting time T esm The time difference is used to obtain the corresponding actual sprint results.

[0070] In step S5, the end face set includes three-dimensional coordinate points on the end face, wherein the end face is also the upper limit face of the crossing range and can be represented as (x, 47, z) for example; in detail, as shown in FIG. Figure 2As shown, if the width of the sprint track is 1.22 meters, for example, the end face set of the sprint track 1 in the three-dimensional coordinate system of the stadium can be expressed as (x0, 47, z0), where the value range of x0 is [0, 5] and the value range of z0 is [0, 1.22]. Since the torso pixel point set and the end face set have a non-empty intersection for the first time, it indicates that the points on the torso area have touched the finish line at this time. Therefore, the first image acquisition time can be used as the actual finishing time T ecm , which can maximize the measurement accuracy of sprint test results and reduce errors, facilitating practical application and promotion.

[0071] Therefore, based on the sprint performance test method described in the aforementioned steps S1 to S5, a new solution for completing the sprint performance test based on a multi-target image segmentation model is provided, namely, the original image data captured in real time by the image acquisition device for multiple contestants are first imported into the pre-trained multi-target image segmentation model in real time to obtain real-time segmentation results of multiple body parts of each contestant, and then for each contestant, the corresponding torso pixel point set is extracted in real time from the real-time segmentation results of the corresponding torso, and if a corresponding torso pixel point set is found to have a non-empty intersection with the corresponding end face set of the sprint track for the first time, the first image acquisition moment corresponding to the torso pixel point set is recorded as the corresponding actual finishing moment, and the time difference between the actual finishing moment and the actual starting moment is calculated to obtain the corresponding actual sprint performance. In this way, the measurement accuracy of the sprint test performance can be maximized and the error can be reduced, which is convenient for practical application and promotion.

[0072] Based on the technical solution of the first aspect, this embodiment further provides a possible design for monitoring false start violations. That is, after obtaining the real-time segmentation results of multiple body parts of each contestant, the method further includes but is not limited to the following steps S311 to S312.

[0073] S311. For each contestant, a corresponding whole-body pixel point set is extracted in real time from the real-time segmentation results of the corresponding multiple body parts, wherein the coordinates of the pixel points in the whole-body pixel point set are three-dimensional coordinates in the three-dimensional coordinate system of the competition field.

[0074] In step S311, since the image acquisition device includes a depth camera for obtaining depth information and / or a three-dimensional camera for capturing high-speed and high-resolution images of the contestants in real time, the original image data and the real-time segmentation results of the multiple body parts will contain depth information and / or three-dimensional coordinate information, and the whole-body pixel point set can be conventionally extracted from the real-time segmentation results of the multiple body parts.

[0075] S312. After the competition notification device issues a start signal, for each competitor, if a corresponding whole-body pixel point set is first found to have a non-empty intersection with the corresponding starting end face set of the sprint track in the three-dimensional coordinate system of the competition field, the second image acquisition time corresponding to the whole-body pixel point set is recorded as the corresponding sensory start time T. fsm , and if the starting time of the sense is determined to be T fsm The actual starting time T esm If the time difference is less than the preset time, the corresponding athlete is judged to have committed a false start violation.

[0076] In step S312, the starting end face set includes three-dimensional coordinate points on the starting end face, wherein the starting end face is also the lower limit end face of the starting range and can be represented as (x, -3, z) for example; in detail, as shown in FIG. Figure 2 As shown, if the width of the sprint track is 1.22 meters, for example, the starting end face set of the sprint track 2 in the three-dimensional coordinate system of the stadium can be expressed as (x1, 47, z1), where the value range of x1 is [0, 5] and the value range of z1 is [1.22, 2.44]. Since the whole body pixel point set and the starting end face set have a non-empty intersection for the first time, it indicates that the points on the athlete's body have touched the starting line at this time. Therefore, the second image acquisition time can be used as the sensory starting time T fsm . Since the track and field competition rules stipulate that 100 / 1000 seconds (0.100 seconds) after the gunshot is used as the critical point for determining a false start, any time when the starting time is less than 100 / 1000 seconds or less than 0.100 seconds from the gunshot is judged as a false start, the preset time length can be specifically designed to be 0.1 seconds. In addition, when it is determined that a contestant has committed a false start violation, the competition notification device can be used to notify all contestants to terminate the competition, or the competition notification device can be used to report the violation of the particular contestant individually to terminate the competition.

[0077] Therefore, based on the aforementioned possible design one, it is possible to monitor the violation of jumping the gun and ensure that the monitoring accuracy is maximized.

[0078] Based on the technical solution of the first aspect mentioned above, this embodiment also provides a possible second design for monitoring lane-crossing violations, that is, after obtaining the real-time segmentation results of multiple body parts of each contestant, the method also includes but is not limited to the following steps S321 to S322.

[0079] S321. For each contestant, a corresponding full-body pixel point set is extracted in real time from the real-time segmentation results of the corresponding multiple body parts, wherein the coordinates of the pixel points in the full-body pixel point set are three-dimensional coordinates in the three-dimensional coordinate system of the competition field.

[0080] S322. After the competition notification device sends a starting signal, for each contestant, if it is first discovered that a corresponding set of full-body pixel points and a corresponding set of boundary surfaces of the sprint track in the three-dimensional coordinate system of the competition venue have a non-empty intersection, then the corresponding contestant is determined to have committed a lane-jumping violation.

[0081] In step S322, the boundary surface set includes three-dimensional coordinate points on the track boundary surface, wherein the track boundary surface specifically includes the track left boundary surface and the track right boundary surface; Figure 2 As shown, referring to the examples of the starting end face and the end face, the right boundary surface set of the sprint track 2 in the three-dimensional coordinate system of the stadium can be expressed as (x1, y1, 1.22), and the left boundary surface set of the sprint track 2 in the three-dimensional coordinate system of the stadium can be expressed as (x1, y1, 2.44), where the value range of x1 is [0, 5] and the value range of y1 is [-3, 47]. Since the whole body pixel point set and the boundary surface set have a non-empty intersection for the first time, it indicates that the points on the player's whole body have touched the track boundary line at this time, so it can be determined that the corresponding player has committed a lane-crossing violation. In addition, when it is determined that a contestant has committed a lane-crossing violation, the competition notification device can be used to notify all contestants to terminate the competition, or the competition notification device can be used to report the violation to the contestant individually to terminate the competition.

[0082] Therefore, based on the aforementioned possible design 2, channel crossing violation monitoring can be carried out and the monitoring accuracy can be maximized.

[0083] Based on the technical solution of the first aspect mentioned above, this embodiment also provides a possible design three for how to provide a finish line warning reminder, that is, after obtaining the real-time segmentation results of multiple body parts of each contestant, the method also includes but is not limited to the following steps S331 to S332.

[0084] S331. For each contestant, a corresponding whole-body pixel point set is extracted in real time from the real-time segmentation results of the corresponding multiple body parts, wherein the coordinates of the pixel points in the whole-body pixel point set are three-dimensional coordinates in the three-dimensional coordinate system of the competition field.

[0085] S332. After the competition notification device sends a starting signal, for each contestant, if a corresponding set of whole-body pixel points is found for the first time to have a non-empty intersection with the lower limit end face set of the finishing range of the corresponding sprint track in the three-dimensional coordinate system of the stadium, the competition notification device is triggered to perform a finishing line warning reminder action.

[0086] In step S332, the set of lower-limit end faces of the finishing range includes three-dimensional coordinate points on the lower-limit end face of the finishing range, where the lower-limit end face of the finishing range is the lower-limit end face of the finishing range. Since the whole-body pixel point set and the set of lower-limit end faces of the finishing range have a non-empty intersection for the first time, indicating that points on the runner's body are about to touch the finish line, the race notification device can be triggered to perform a finish line warning. Furthermore, the finish line warning warning can be, but is not limited to, a voice reminder.

[0087] Therefore, based on the aforementioned possible design three, a finish line warning reminder can be issued and the accuracy of the warning monitoring can be maximized.

[0088] Based on the technical solution of the first aspect mentioned above, this embodiment further provides a fourth possible design for reconstructing a three-dimensional model of the contestants within the starting range and / or the finishing range. That is, after receiving in real time the original image data captured by the image acquisition device for the multiple contestants in real time, the method further includes but is not limited to the following steps S301 to S304.

[0089] S301. For any contestant among the multiple contestants, record an image sequence of the corresponding contestant within the starting range and / or the finishing range, wherein the image sequence includes a plurality of image groups that are consecutive in image acquisition timing, and the image group includes a plurality of frames of images acquired by different plurality of image acquisition devices at the same image acquisition moment.

[0090] S302. For each frame of the multiple frames, the position of any contestant at the corresponding image acquisition moment is calculated based on a set of whole-body pixel points corresponding to and extracted from the real-time segmentation results of multiple body parts of any contestant, and the depth data of the corresponding image acquisition device is also integrated to generate a three-dimensional point cloud of the any contestant at the corresponding image acquisition moment from the perspective of the corresponding image acquisition device.

[0091] In step S302, the location can be obtained based on the center point of the whole-body pixel point set; the specific fusion generation process of the three-dimensional point cloud is an existing technical means and will not be repeated here.

[0092] S303. For each image group in the image sequence, use multi-view stereo vision technology to fuse the three-dimensional point clouds of any contestant at the corresponding image acquisition moment under different perspectives, and generate a three-dimensional model of any contestant at the corresponding image acquisition moment through triangulation.

[0093] In step S303, the multi-view stereo vision technology is a technology for reconstructing three-dimensional geometric shapes from images from multiple perspectives. It uses a stereo matching algorithm and images from different perspectives to restore the three-dimensional structure of the scene. Therefore, the specific process of the fusion processing and the triangulation is an existing technical means and will not be repeated here.

[0094] S304. Summarize the location and 3D model of any contestant at each image acquisition moment to obtain a 3D model reconstruction result of any contestant within the starting range and / or the finishing range.

[0095] After step S304, the 3D visualization tool OpenGL can also be used to display the 3D model reconstruction results of any contestant in the starting range and / or the finishing range, so as to conduct a 3D start / finish review of any contestant after the race. For example, it allows the user to observe from any angle and use the contestant's depth information to record the contestant's 3D motion trajectory and dynamically display it. In addition, the number plate list of any contestant, the actual starting time T esm , actual finishing time T ecm The actual sprint results are also recorded to serve as the sprint test result data for the corresponding athletes.

[0096] Based on the aforementioned possible design four, a three-dimensional model of the contestant within the starting range and / or the finishing range can also be reconstructed to facilitate a three-dimensional starting / finishing range review after the race, which is beneficial for helping the coaching system train the contestants' weaknesses and further improve practicality.

[0097] A second aspect of this embodiment provides a virtual system for implementing the sprint performance testing method described in the first aspect or any possible design of the first aspect, adapted to be disposed in a computer device communicatively connected to a competition notification device and an image acquisition device, respectively, and comprising a correspondence acquisition unit, an image segmentation processing unit, a torso pixel extraction unit, a start time recording unit, and a sprint performance measurement unit.

[0098] The correspondence obtaining unit is configured to obtain a one-to-one correspondence between a plurality of sprint tracks and a plurality of contestants, wherein the plurality of sprint tracks are adjacent to each other and constitute a sprint stadium;

[0099] The image segmentation processing unit is configured to receive, in real time, raw image data captured by the image acquisition device of the multiple contestants in real time after the multiple contestants enter the pre-starting position of the sprint stadium, and import the raw image data into a pre-trained multi-objective image segmentation model in real time to obtain real-time segmentation results of multiple body parts of each of the multiple contestants, wherein the image acquisition device includes a depth camera for obtaining depth information and / or a three-dimensional camera for capturing high-speed and high-resolution images of the contestants in real time, and the multiple body parts include a head, a torso, and a limb.

[0100] The torso pixel point extraction unit is communicatively connected to the image segmentation processing unit and is configured to extract, in real time, a corresponding torso pixel point set from the real-time segmentation results of the corresponding torso for each contestant, wherein the coordinates of the pixel points in the torso pixel point set are three-dimensional coordinates in a three-dimensional coordinate system of the competition field, wherein the X-axis direction of the three-dimensional coordinate system of the competition field is perpendicular to the sprint track, the Y-axis direction of the three-dimensional coordinate system of the competition field is the direction from the starting point to the finish line, and the Z-axis direction of the three-dimensional coordinate system of the competition field is perpendicular to the boundary line of the sprint track and passes through the starting point;

[0101] The starting time recording unit is configured to record the time when the competition notification device sends the starting signal as the actual starting time;

[0102] The sprint performance measurement unit is communicatively connected to the corresponding relationship acquisition unit, the torso pixel point extraction unit, and the starting time recording unit, and is configured to, after the competition notification device issues a starting signal, for each competitor, if a non-empty intersection is first found between a corresponding torso pixel point set and a corresponding end face set of the sprint track in the three-dimensional coordinate system of the competition field, record the first image acquisition time corresponding to the torso pixel point set as the corresponding actual finishing time T ecm , and calculate the actual finishing time T ecm The actual starting time T esm The time difference is used to obtain the corresponding actual sprint results.

[0103] The working process, working details and technical effects of the aforementioned system provided in the second aspect of this embodiment can be referred to the first aspect or any possible design of the short-distance running performance testing method described in the first aspect, and will not be repeated here.

[0104] A third aspect of this embodiment provides a physical system for implementing the sprint performance testing method described in the first aspect or any possible design of the first aspect, including a competition notification device, an image acquisition device, and a computer device;

[0105] The race notification device is communicatively connected to the computer device, and is used to send a start signal and trigger the computer device to record the actual start time;

[0106] The image acquisition device is communicatively connected to the computer device and is used to capture raw image data of multiple contestants in real time and transmit the captured results to the computer device in real time;

[0107] The computer device is used to execute the sprint performance testing method as described in the first aspect or any possible design of the first aspect.

[0108] The working process, working details and technical effects of the aforementioned system provided in the third aspect of this embodiment can be referred to the first aspect or any possible design of the short-distance running performance testing method described in the first aspect, and will not be repeated here.

[0109] like Figure 3 As shown, the present embodiment fourth aspect provides a kind of computer equipment that performs the sprint performance test method described in any possible design as in the first aspect or the first aspect, including a memory, a processor and a transceiver that are connected in sequence, wherein the memory is used to store a computer program, the transceiver is used to send and receive messages, and the processor is used to read the computer program, and performs the sprint performance test method described in any possible design as in the first aspect or the first aspect. Specifically, the memory may include, but is not limited to, a random access memory (Random-Access Memory, RAM), a read-only memory (Read-Only Memory, ROM), a flash memory (Flash Memory), a first-in-first-out memory (First Input First Output, FIFO) and / or a first-in-last-out memory (First Input Last Output, FILO) and the like; the processor may, but is not limited to, adopt a microprocessor of the STM32F105 series. In addition, the computer equipment may also include, but is not limited to, a power module, a display screen and other necessary components.

[0110] The working process, working details and technical effects of the aforementioned computer device provided in the fourth aspect of this embodiment can be referred to the short-distance running performance testing method described in the first aspect or any possible design of the first aspect, and will not be repeated here.

[0111] A fourth aspect of this embodiment provides a computer-readable storage medium storing instructions for a sprint performance testing method as described in the first aspect or any possible design of the first aspect. Specifically, the computer-readable storage medium stores instructions that, when executed on a computer, execute the sprint performance testing method as described in the first aspect or any possible design of the first aspect. The computer-readable storage medium refers to a data storage medium and may include, but is not limited to, a floppy disk, an optical disk, a hard disk, a flash memory, a USB flash drive, and / or a memory stick. The computer may be a general-purpose computer, a dedicated computer, a computer network, or other programmable device.

[0112] The working process, working details and technical effects of the aforementioned computer-readable storage medium provided in the fourth aspect of this embodiment can be referred to the sprint performance testing method described in the first aspect or any possible design of the first aspect, and will not be repeated here.

[0113] A fifth aspect of this embodiment provides a computer program product, including a computer program or instructions, which, when executed by a computer, implements the sprint performance testing method described in the first aspect or any possible design of the first aspect. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device.

[0114] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of protection of the present invention.

Claims

1. A sprint performance test method based on image segmentation, characterized in that: The method is executed by a computer device that is communicatively connected to the competition notification device and the image acquisition device, and includes: Obtaining a one-to-one correspondence between a plurality of sprint tracks and a plurality of contestants, wherein the plurality of sprint tracks are sequentially adjacent and constitute a sprint stadium; After the multiple contestants enter the pre-starting position of the sprint stadium, raw image data captured by the image acquisition device for the multiple contestants are received in real time, and the raw image data are imported into a pre-trained multi-objective image segmentation model in real time to obtain real-time segmentation results of multiple body parts of each of the multiple contestants, wherein the image acquisition device includes a depth camera for obtaining depth information and / or a three-dimensional camera for capturing high-speed and high-resolution images of the contestants in real time, and the multiple body parts include a head, a torso, and a limb. For each of the contestants, a corresponding torso pixel point set is extracted in real time from the real-time segmentation results of the corresponding torso, wherein the coordinates of the pixel points in the torso pixel point set are three-dimensional coordinates in a three-dimensional coordinate system of the competition field, wherein the X-axis direction of the three-dimensional competition field coordinate system is perpendicular to the sprint track, the Y-axis direction of the three-dimensional competition field coordinate system is the direction from the starting point to the finish line, and the Z-axis direction of the three-dimensional competition field coordinate system is perpendicular to the boundary line of the sprint track and passes through the starting point; When the competition notification device issues a start signal, recording the time when the start signal is issued as the actual start time; After the competition notification device sends the starting signal, for each competitor, if a non-empty intersection is first found between a corresponding torso pixel point set and a corresponding end face set of the sprint track in the three-dimensional coordinate system of the competition field, the first image acquisition time corresponding to the torso pixel point set is recorded as the corresponding actual finishing time T ecm , and calculate the actual finishing time T ecm The actual starting time T esm The time difference is used to obtain the corresponding actual sprint results.

2. The sprint performance testing method according to claim 1, wherein: The image acquisition device includes three depth cameras for obtaining depth information and two three-dimensional cameras for capturing high-speed and high-resolution images of contestants in real time, wherein the first depth camera is arranged on the upper end face of the starting range of the sprint stadium with its lens facing the sprint stadium, the second depth camera is arranged above the main running range of the sprint stadium with its lens facing the sprint stadium, and the third depth camera is arranged on the lower end face of the finishing range of the sprint stadium with its lens facing the sprint stadium. The first three-dimensional camera is arranged on the lower end face of the starting range with its lens facing the sprint stadium, and the second three-dimensional camera is arranged on the upper end face of the finishing range with its lens facing the sprint stadium. The main running range is located between the starting range and the finishing range.

3. The sprint performance testing method according to claim 2, characterized in that: After receiving in real time the raw image data of the multiple contestants captured by the image acquisition device respectively in real time, the method further includes: For any one of the plurality of contestants, recording an image sequence of the corresponding contestant within the starting range and / or the finishing range, wherein the image sequence comprises a plurality of image groups that are sequentially sequential in image acquisition time sequence, and the image groups comprise a plurality of image frames acquired by different plurality of image acquisition devices at the same image acquisition time; For each frame of the multiple images, calculate the location of the contestant at the corresponding image capture moment based on a corresponding set of full-body pixel points extracted from the real-time segmentation results of multiple body parts of the contestant, and further fuse the depth data of the corresponding image capture device to generate a three-dimensional point cloud of the contestant at the corresponding image capture moment from the perspective of the corresponding image capture device; For each image group in the image sequence, using multi-view stereo vision technology to fuse the three-dimensional point clouds of the contestant at the corresponding image capture moment from different perspectives, and generate a three-dimensional model of the contestant at the corresponding image capture moment through triangulation; The position and three-dimensional model of any contestant at each image acquisition moment are summarized to obtain a three-dimensional model reconstruction result of any contestant in the starting range and / or the finishing range.

4. The sprint performance testing method according to claim 3, wherein: After obtaining the three-dimensional model reconstruction result of any contestant within the starting range and / or the finishing range, the method further includes: The 3D visualization tool OpenGL is used to display the three-dimensional model reconstruction result of any contestant in the starting range and / or the finishing range, so as to conduct a three-dimensional starting / finishing review of any contestant after the race.

5. The sprint performance testing method according to claim 1, wherein: The multi-object image segmentation model is obtained based on the pre-training of the Mask R-CNN model.

6. The sprint performance testing method according to claim 1, wherein: After obtaining the real-time segmentation results of the multiple body parts of each contestant, the method further includes: For each contestant, extracting in real time a corresponding whole-body pixel point set from the real-time segmentation results of the corresponding multiple body parts, wherein the coordinates of the pixel points in the whole-body pixel point set are three-dimensional coordinates in the three-dimensional coordinate system of the competition field; After the competition notification device sends a start signal, for each contestant, if a corresponding whole-body pixel point set is first found to have a non-empty intersection with the starting end face set of the corresponding sprint track in the three-dimensional coordinate system of the competition field, the second image acquisition time corresponding to the whole-body pixel point set is recorded as the corresponding sensory start time T fsm , and if the starting time of the sense is determined to be T fsm The actual starting time T esm If the time difference is less than the preset time, the corresponding athlete is judged to have committed a false start violation.

7. The sprint performance testing method according to claim 1, characterized in that: After obtaining the real-time segmentation results of the multiple body parts of each contestant, the method further includes: For each contestant, extracting in real time a corresponding whole-body pixel point set from the real-time segmentation results of the corresponding multiple body parts, wherein the coordinates of the pixel points in the whole-body pixel point set are three-dimensional coordinates in the three-dimensional coordinate system of the competition field; After the competition notification device sends a starting signal, for each contestant, if a corresponding set of whole-body pixel points is found for the first time to have a non-empty intersection with the boundary surface set of the corresponding sprint track in the three-dimensional coordinate system of the stadium, it is determined that the corresponding contestant has committed a lane-crossing violation.

8. The sprint performance testing method according to claim 1, wherein: After obtaining the real-time segmentation results of the multiple body parts of each contestant, the method further includes: For each contestant, extracting in real time a corresponding whole-body pixel point set from the real-time segmentation results of the corresponding multiple body parts, wherein the coordinates of the pixel points in the whole-body pixel point set are three-dimensional coordinates in the three-dimensional coordinate system of the competition field; After the competition notification device sends a starting signal, for each contestant, if a corresponding set of whole-body pixel points is found for the first time to have a non-empty intersection with the lower limit end face set of the finishing range of the corresponding sprint track in the three-dimensional coordinate system of the stadium, the competition notification device is triggered to perform a finishing line warning reminder action.

9. A sprint performance testing system based on image segmentation, characterized in that: Suitable for being arranged in a computer device that is communicatively connected to a competition notification device and an image acquisition device, respectively, and includes a correspondence acquisition unit, an image segmentation processing unit, a torso pixel point extraction unit, a start time recording unit, and a sprint performance measurement unit; The correspondence obtaining unit is configured to obtain a one-to-one correspondence between a plurality of sprint tracks and a plurality of contestants, wherein the plurality of sprint tracks are adjacent to each other and constitute a sprint stadium; The image segmentation processing unit is configured to receive, in real time, raw image data captured by the image acquisition device of the multiple contestants in real time after the multiple contestants enter the pre-starting position of the sprint stadium, and import the raw image data into a pre-trained multi-objective image segmentation model in real time to obtain real-time segmentation results of multiple body parts of each of the multiple contestants, wherein the image acquisition device includes a depth camera for obtaining depth information and / or a three-dimensional camera for capturing high-speed and high-resolution images of the contestants in real time, and the multiple body parts include a head, a torso, and a limb. The torso pixel point extraction unit is communicatively connected to the image segmentation processing unit and is configured to extract, in real time, a corresponding torso pixel point set from the real-time segmentation results of the corresponding torso for each contestant, wherein the coordinates of the pixel points in the torso pixel point set are three-dimensional coordinates in a three-dimensional coordinate system of the competition field, wherein the X-axis direction of the three-dimensional coordinate system of the competition field is perpendicular to the sprint track, the Y-axis direction of the three-dimensional coordinate system of the competition field is the direction from the starting point to the finish line, and the Z-axis direction of the three-dimensional coordinate system of the competition field is perpendicular to the boundary line of the sprint track and passes through the starting point; The starting time recording unit is configured to record the time when the competition notification device sends the starting signal as the actual starting time; The sprint performance measurement unit is communicatively connected to the corresponding relationship acquisition unit, the torso pixel point extraction unit, and the starting time recording unit, and is configured to, after the competition notification device issues a starting signal, for each competitor, if a non-empty intersection is first found between a corresponding torso pixel point set and a corresponding end face set of the sprint track in the three-dimensional coordinate system of the competition field, record the first image acquisition time corresponding to the torso pixel point set as the corresponding actual finishing time T ecm , and calculate the actual finishing time T ecm The actual starting time T esm The time difference is used to obtain the corresponding actual sprint results.

10. A sprint performance testing system based on image segmentation, characterized in that: Including competition notification equipment, image acquisition equipment and computer equipment; The race notification device is communicatively connected to the computer device, and is used to send a start signal and trigger the computer device to record the actual start time; The image acquisition device is communicatively connected to the computer device and is used to capture raw image data of multiple contestants in real time and transmit the captured results to the computer device in real time; The computer device is used to execute the sprint performance testing method according to any one of claims 1 to 8.

Images