Method and device for detecting long jump and computer readable medium

Abstract

The invention discloses a method and a device for detecting a long jump and a computer readable medium. The method comprises the following steps: acquiring a video of the long jump; extracting a key frame in the video, wherein the key frame comprises a take-off frame and a landing frame; identifying a first key point in the take-off frame and a second key point in the landing frame, and obtaining a first key point coordinate corresponding to the first key point and a second key point coordinate corresponding to the second key point; according to the scale line coordinates and the first key point coordinates of the long jump mat in the key frame, whether the long jump mat is qualified or not is determined; under the condition that the long jump is qualified, determining an initial long jump distance according to the scale line coordinates and the second key point coordinates; and performing error correction on the initial long jump distance to obtain the long jump distance. According to the method for measuring the long jump, the error of measuring the long jump distance can be reduced, so that the accuracy of measuring the long jump is improved. Moreover, manual participation can be reduced, so that the process of measuring the long jump is simplified, and the efficiency of measuring the long jump is improved.

Description

Technical Field

[0001] The present disclosure generally relates to the field of image processing technology. More specifically, the present disclosure relates to a method and apparatus for detecting long jump and a computer-readable medium. Background Art

[0002] Long jump is one of the main events in physical education and track and field competitions. Long jump is usually divided into standing long jump, running long jump and pole long jump. The long jump result is usually one of the important indicators to measure a person's physical fitness.

[0003] In the prior art, the long jump distance is usually measured manually. Specifically, the measurement personnel use a tape measure or other tool to measure the distance from the starting point to the landing point. This measurement method requires manual measurement step by step when measuring the long jump distance of each person. Therefore, this measurement method has a large workload and takes a long time, resulting in low measurement efficiency. Moreover, this measurement method may result in errors, resulting in inaccurate measurements.

[0004] In view of this, there is an urgent need to provide a solution for detecting long jump, so as to improve the accuracy of measuring long jump and improve the efficiency of measurement. Summary of the invention

[0005] In order to at least solve one or more technical problems mentioned above, the present disclosure proposes a method and an apparatus for detecting long jump and a computer-readable medium in multiple aspects.

[0006] In a first aspect, the present disclosure provides a method for detecting long jump, comprising: obtaining a video of long jump; extracting key frames in the video, the key frames including a take-off frame and a landing frame; identifying a first key point in the take-off frame and a second key point in the landing frame, and obtaining a first key point coordinate corresponding to the first key point and a second key point coordinate corresponding to the second key point; determining whether the long jump is qualified based on the scale line coordinates of the long jump mat in the key frame and the first key point coordinates; if the long jump is qualified, determining an initial long jump distance based on the scale line coordinates and the second key point coordinates; and performing error correction on the initial long jump distance to obtain a long jump distance.

[0007] In a second aspect, the present disclosure provides a device for detecting long jump, comprising: a video acquisition module for acquiring a video of the long jump; a key frame extraction module for extracting key frames in the video, the key frames including a take-off frame and a landing frame; a key point identification module for identifying a first key point in the take-off frame and a second key point in the landing frame, and obtaining a first key point coordinate corresponding to the first key point and a second key point coordinate corresponding to the second key point; a long jump qualification determination module for determining whether the long jump is qualified based on the scale line coordinates of the long jump mat in the key frame and the first key point coordinates; a distance determination module for determining an initial long jump distance based on the scale line coordinates and the second key point coordinates if the long jump is qualified; and an error correction module for performing error correction on the initial long jump distance to obtain a long jump distance.

[0008] In a third aspect, the present disclosure provides a computer-readable medium having a computer program stored thereon, and when the computer program is executed by a processor, the processor executes the method for detecting long jump provided in the first aspect.

[0009] By the method for detecting long jump provided above, the disclosed embodiment first extracts key frames in the long jump video; then, identifies key points in the key frames and extracts key coordinates of the key points; then, according to the scale line coordinates and key point coordinates of the long jump mat in the key frames, it is possible to determine whether the long jump is qualified. And, in the case of qualified long jump, the initial long jump distance is determined according to the scale line coordinates and key point coordinates; finally, the initial long jump distance is error corrected to obtain the long jump distance, which can reduce the error of measuring the long jump distance, thereby improving the accuracy of measuring the long jump. Moreover, the above-mentioned method for measuring long jump can reduce manual participation, thereby simplifying the process of measuring long jump, so as to improve the efficiency of measuring long jump. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] By reading the detailed description below with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present disclosure will become readily understood. In the accompanying drawings, several embodiments of the present disclosure are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0011] Figure 1 An exemplary flow chart 100 showing a method for detecting a long jump according to some embodiments of the present disclosure;

[0012] Figure 2 An exemplary flowchart 120 showing a method of extracting key frames in a video according to some embodiments of the present disclosure;

[0013] Figure 3An exemplary flowchart 130 showing a method of identifying key points in a key frame and obtaining key point coordinates according to some embodiments of the present disclosure;

[0014] Figure 4 An exemplary flow chart 140 showing a method of determining whether a long jump is qualified according to some embodiments of the present disclosure;

[0015] Figure 5 An exemplary flow chart 150 showing a method of determining an initial long jump distance according to some embodiments of the present disclosure;

[0016] Figure 6 An exemplary flow chart 157 showing a method of determining an initial long jump distance according to some embodiments of the present disclosure;

[0017] Figure 7 An exemplary flow chart 158 ​​showing a method of determining an initial long jump distance according to some other embodiments of the present disclosure;

[0018] Figure 8 An exemplary flow chart 160 showing a method of performing error correction on an initial long jump distance according to some embodiments of the present disclosure;

[0019] Fig. 9 An exemplary structural block diagram of an apparatus 900 for detecting long jump according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present disclosure.

[0021] It should be understood that the terms "include" and "comprising" used in the specification and claims of the present disclosure indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0022] It should also be understood that the terms used in this disclosure are only for the purpose of describing specific embodiments and are not intended to limit the disclosure. As used in this disclosure and claims, the singular forms of "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise. It should also be further understood that the term "and / or" used in this disclosure and claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations.

[0023] As used in this specification and claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0024] The specific implementation of the present disclosure is described in detail below with reference to the accompanying drawings.

[0025] Exemplary application scenarios.

[0026] In the standing long jump teaching of a large number of students, it is necessary to measure the long jump of each student to obtain the long jump score of each student. For the convenience of description, the person who performs the long jump is called a long jumper.

[0027] An existing solution for measuring long jump specifically includes: the surveyor measures the distance from the starting point to the landing point using a tape measure or other tool. Since the surveyor needs to use a tape measure to measure the long jump distance of each long jumper, and the surveyor needs to read and record the distance of each long jump, this measurement solution has a large workload and leads to low measurement efficiency. Moreover, the surveyor may make errors when measuring the distance from the starting point to the landing point using a tape measure or other tool, resulting in inaccurate measurement.

[0028] Another existing solution for measuring long jump usually uses a detection device based on a sensor such as infrared or laser radar to measure the distance of the long jump, but the above detection device is usually expensive and has a high degree of complexity. Therefore, a new method for measuring long jump is needed to improve the efficiency and accuracy of measuring long jump and reduce the measurement cost. The specific embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.

[0029] Exemplary application scenarios.

[0030] In view of this, the disclosed embodiment provides a method for detecting long jump, which first extracts key frames in a long jump video; then, identifies key points in the key frames and extracts key coordinates of the key points; then, determines whether the long jump is qualified based on the scale line coordinates and key point coordinates of the long jump mat in the key frames, and, if the long jump is qualified, determines the initial long jump distance based on the scale line coordinates and the key point coordinates; finally, performs error correction on the initial long jump distance to obtain the long jump distance, which can reduce manual participation, thereby simplifying the process of measuring long jump, so as to improve the efficiency of measuring long jump; moreover, it can determine whether the long jump is qualified, and at the same time, it can reduce the error of measuring the long jump distance, thereby improving the accuracy of measuring long jump.

[0031] Figure 1 An exemplary flow chart 100 of a method for detecting a long jump according to some embodiments of the present disclosure is shown.

[0032] As shown in the figure, in step S110, a video of long jump is obtained; in step S120, key frames in the video are extracted, and the key frames include a take-off frame and a landing frame; in step S130, a first key point in the take-off frame and a second key point in the landing frame are identified, and a first key point coordinate corresponding to the first key point and a second key point coordinate corresponding to the second key point are obtained; in step S140, whether the long jump is qualified is determined based on the scale line coordinates and the first key point coordinates of the long jump mat in the key frame; in step S150, if the long jump is qualified, the initial long jump distance is determined based on the scale line coordinates and the second key point coordinates; in step S160, the initial long jump distance is error corrected to obtain the long jump distance.

[0033] In the embodiment of the present disclosure, the "long jump video" in step S110 refers to a video of the entire process of a person's long jump taken by a camera. The camera is set on either side of the long jump mat, facing the side of the person performing the long jump. Moreover, the position and shooting direction of the camera need to be fixed to avoid image shaking.

[0034] In the embodiment disclosed herein, there is a rectangular long jump mat in the long jump venue, and the "long jump mat" here refers to a mat for long jump, on which people perform long jump. There is a distance coordinate axis on the rectangular long jump mat, which is parallel to the long side of the rectangular long jump mat, and there are multiple scale lines on the distance coordinate axis, which are used to measure the distance between the landing point and the take-off line, each scale line is perpendicular to the distance coordinate axis, and each scale line has a corresponding distance value, and the distance value represents the distance between the scale line corresponding to the distance value and the take-off line in real space. For example, if the distance value corresponding to the scale line a is 1 meter, then the distance between the scale line a and the take-off line in real space is 1 meter. The distance value corresponding to the scale line can include scale lines corresponding to 1 meter, 1.5 meters and 2 meters. In this embodiment, the number of scale lines can be set voluntarily. Moreover, the distance between any two adjacent scale lines on the long jump mat is the same.

[0035] The above-mentioned "landing point" refers to the point where the long jumper touches the ground and is closest to the take-off line when landing. For example, after a long jumper performs a complete and standard long jump action, the landing point of the long jumper is the left toe or the right toe. It should be noted that the above-mentioned long jump mat also has a take-off line, which is the starting line for the long jumper to take off, and the distance between the take-off line and the landing point is the long jump distance of the long jumper.

[0036] The above-mentioned long jump video includes one or more frames of images. For the convenience of description, each frame of images included in the above-mentioned long jump video is referred to as a "long jump frame". The "key frame" in step S120 refers to the long jump frame corresponding to one or more key actions of the human body in the long jump process included in the long jump frame, and the "key action" here refers to the take-off action, vacating and landing. The key frame includes a take-off frame and a landing frame. The above-mentioned "take-off frame" refers to the long jump frame corresponding to the long jump personnel when performing the preparatory action or pre-action of the take-off, and the above-mentioned "landing frame" refers to the long jump frame corresponding to the person after the person lands stably. The stable landing here refers to the person landing after the landing, and the landing point closest to the take-off line is stable and unchanged.

[0037] In some embodiments, the key frame also includes a flying frame, where the "flying frame" refers to one or more frames of images corresponding to a person performing a long jump in the air in a long jump video.

[0038] In some embodiments of the present disclosure, it is necessary to first divide the long jump video into one or more long jump frames, and then identify and extract key frames from the long jump frames.

[0039] In some embodiments, key points refer to the key parts or joints of the long jump personnel when participating in the long jump, which are used to describe the key parts of the long jump personnel's body or joints. The key points may include the head, left shoulder, right shoulder, left ankle and right ankle of the long jump personnel. In some embodiments of the present disclosure, for the convenience of description, the key points identified in the take-off frame are referred to as the first key points, and the key points identified in the landing frame are referred to as the second key points. Moreover, the first key point is the same as the human body part of the second key point. For example, the first key point can be the head, left shoulder, right shoulder, left elbow, right elbow, left ankle and right ankle of the long jump personnel in the take-off frame, and the second key point is the head, left shoulder, right shoulder, left elbow, right elbow, left ankle and right ankle of the long jump personnel in the landing frame.

[0040] In step S130, after identifying the first key point in the take-off frame and the second key point in the landing frame, it is also necessary to obtain the coordinates corresponding to the first key point and the coordinates corresponding to the second key point. The coordinates corresponding to the first key point refer to the position of the first key point in the take-off frame. Similarly, the coordinates corresponding to the second key point refer to the position of the first key point in the landing frame. For ease of description, the "coordinates corresponding to the first key point" are referred to as the first key point coordinates, and the "coordinates corresponding to the second key point" are referred to as the second key point coordinates.

[0041] Specifically, the coordinates of the first key point can be represented by the position of the first key point corresponding to the pixel in the jump frame, for example, the coordinates of the first key point can be the pixel of the 2nd row and the 3rd column, and can also be represented by the physical unit corresponding to the first key point in the jump frame, for example, the coordinates of the first key point can be the position with a horizontal coordinate of 4 mm and a vertical coordinate of 3 mm in the jump frame. The representation method of the coordinates of the first key point is consistent with the representation method of the coordinates of the second key point.

[0042] The "scale line coordinates" in step S140 refer to the coordinates corresponding to the straight line where the scale lines of the long jump mat are located in the long jump frame.

[0043] As can be known from the description of the foregoing embodiment, there is a take-off line on the long jump mat. In the present embodiment, the take-off line is a part of the scale lines of the long jump mat. The first key point also includes the above-mentioned landing point and the take-off point. The landing point can be the left toe or the right toe. The above-mentioned take-off point refers to the point that the long jumper is closest to the take-off line when preparing to take off and is in contact with the long jump mat or the ground. For example, the take-off point can be the left toe or the right toe. The first key point coordinates include the coordinates corresponding to the landing point and the coordinates corresponding to the take-off point. Here, the "coordinates corresponding to the landing point" are referred to as the landing point coordinates, and the "coordinates corresponding to the take-off point" are referred to as the take-off point coordinates. The scale line coordinates of the long jump mat also include the coordinates corresponding to the straight line where the take-off line is located in the long jump frame, and the "straight line where the take-off line is located in the long jump frame" is referred to as the take-off line coordinates.

[0044] In the above step S140, according to the scale line coordinates of the long jump mat in the key frame and the first key point coordinates, determining whether the long jump is qualified specifically includes: first, according to the take-off line coordinates and the take-off point coordinates, determining the relationship between the take-off line and the take-off point. Then, judging whether the take-off point exceeds the take-off line by the relationship between the take-off line and the take-off point. If the take-off point exceeds the take-off line, it is determined that the long jump is unqualified; if the take-off point does not exceed the take-off line, it is determined that the long jump is qualified.

[0045] In the embodiment of the present disclosure, there are multiple scale lines on the long jump mat on one side of the take-off line, the direction from the take-off line to the multiple scale lines is the long jump direction, and there are no scale lines on the long jump mat on the other side of the take-off line. Taking the take-off line as the dividing point, the side of the take-off line close to the multiple scale lines is called the "valid performance side", and the side of the take-off line away from the multiple scale lines is called the "invalid performance side". The above-mentioned "taking-off point exceeds the take-off line" means that the take-off point is located on the valid performance side, for example, the take-off point is located on the scale line corresponding to the distance value of 0.5 meters. And "the take-off point does not exceed the take-off line" means that the take-off point is located on the invalid performance side. For example, if the distance between a take-off point and the take-off line is 0.3 meters, and the distance between the take-off point and the scale line corresponding to the distance value of 0.5 meters is 0.8 meters, then the take-off point does not exceed the take-off line.

[0046] In this embodiment, the positional relationship between the take-off line and the take-off point can be determined through the take-off line coordinates and the take-off point coordinates to determine whether the long jump is qualified.

[0047] If the long jump is determined to be qualified, step S150 is executed, which specifically includes: determining the initial long jump distance according to the scale line coordinates and the second key point coordinates, that is, determining the scale line corresponding to the second key point, where the "scale line corresponding to the second key point" is referred to as the landing scale line. Next, the distance between the landing scale line and the take-off line is calculated, where the "distance between the landing scale line and the take-off line" is referred to as the initial long jump distance.

[0048] In the event of a failure in the long jump, the long jump result of the long jumper will be determined to be invalid.

[0049] Further, after the initial long jump distance is obtained in step S150, step S160 is executed. Since the obtained initial long jump distance has an error, the initial long jump distance needs to be corrected in step S160 to obtain a corrected initial long jump distance, that is, the above-mentioned long jump distance.

[0050] It can be known from the description of the above embodiments that a method for detecting long jump provided by the disclosed embodiment can determine whether the long jump is qualified by the scale line coordinates and key point coordinates of the long jump mat in the key frame. Moreover, when the long jump is qualified, the initial long jump distance is determined according to the scale line coordinates and the key point coordinates, and the initial long jump distance is error corrected to obtain the long jump distance, so that the error of measuring the long jump distance can be reduced, thereby improving the accuracy of measuring the long jump. At the same time, the method can reduce manual participation, thereby simplifying the process of measuring the long jump, so as to improve the efficiency of measuring the long jump.

[0051] In some embodiments, step S110 specifically includes: setting the position of the camera device, calibrating the camera device to calibrate the lens of the camera device, and then using the calibrated camera device to shoot the complete process of a person's long jump, thereby obtaining a long jump video.

[0052] In order to determine the mapping relationship between the three-dimensional geometric position of each point of a real-space object and the corresponding points of the real-space object in the captured two-dimensional image, a geometric model of camera imaging is established. The parameters of these geometric models are camera parameters. Camera parameters may include internal and external parameters and distortion parameters, etc. The internal parameters of the camera may include the focal length, optical center and radial distortion coefficient of the lens. The external parameters of the camera may include the orientation of the camera relative to certain world coordinate systems. The above-mentioned "camera calibration" refers to the process of solving camera parameters through experiments and calculations.

[0053] In some embodiments, the camera calibration method includes a chessboard calibration method. Specifically, the process of the chessboard calibration method includes: setting a chessboard, which is composed of a plurality of black and white squares, and determining the coordinates of the vertices of each square in the chessboard; then, taking a plurality of photos of the chessboard through the above-mentioned camera; in this embodiment, the number of photos taken of the chessboard can be set by oneself; then, through opencv (Open Source Computer Vision Library, open source computer vision library) according to the coordinates of the vertices of each square in the chessboard, the plurality of photos of the chessboard are calibrated, that is, the distortion of the photo is corrected according to the internal and external parameters and distortion parameters of the camera. The "distortion" here refers to the abnormality of the object in the captured image. For example, when shooting a straight pole, the image captured is a bent pole. This is because the shape of the object in the captured image has changed due to the lens of the camera.

[0054] The intrinsic and extrinsic parameters and distortion parameters after the distortion correction are used as the final camera parameters of the camera. Those skilled in the art can understand that the camera calibration can be realized by calling the Zhang Zhengyou calibration function by opencv. The Zhang Zhengyou calibration function here refers to the function used to realize the camera calibration in opencv. The specific process of the camera calibration is not repeated here.

[0055] In some embodiments, the camera calibration method further includes a zero-distortion camera calibration method and a self-calibration algorithm. Those skilled in the art will appreciate that they may select any one of the above camera calibration methods.

[0056] By shooting a long jump video with a camera after camera calibration, it is possible to avoid distortion or deformation of the image captured by the camera, thereby avoiding the inability to accurately determine the distance between any two points in the image based on the image captured by the camera.

[0057] Figure 2 An exemplary flow chart 120 of a method for extracting key frames in a video according to some embodiments of the present disclosure is shown. It can be understood that the method for extracting key frames in a video is a specific implementation of the aforementioned step S120, so the aforementioned method is combined with the method described in the embodiment of the present disclosure. Figure 1 and Figure 2 The features described can analogously apply here.

[0058] As shown in the figure, in step S121, a long jump waiting area is set in the long jump frame, and the long jump waiting area is located in the long jump waiting area of ​​the long jump mat; in step S122, the intersection and union ratio of the long jump personnel and the long jump waiting area in the long jump frame is determined to obtain a first intersection and union ratio; in step S123, the intersection and union ratio between each two adjacent long jump frames is determined to obtain a second intersection and union ratio; in step S124, in response to the first intersection and union ratio being less than a preset first threshold value, the long jump frame corresponding to the first intersection and union ratio is used as a take-off frame; in step S125, in response to the second intersection and union ratio being greater than a preset second threshold value, the long jump frame corresponding to the second intersection and union ratio is used as a landing frame.

[0059] Here, the "long jump waiting area" refers to the area where the long jumper is located when the long jumper is in the preparation stage of the long jump, and the long jump waiting area includes various parts of the long jumper's body in the long jump frame. In step S121, the area near the take-off line of the long jump mat in the long jump frame can be set as the long jump waiting area. Those skilled in the art will appreciate that the long jump waiting area in the present embodiment can be set voluntarily without any limitation.

[0060] Then, in step S122, the intersection-and-combination ratio between the area corresponding to the long jumper and the long jump waiting area in each long jump frame is determined, and the "intersection-and-combination ratio between the area corresponding to the long jumper and the long jump waiting area" is referred to as the first intersection-and-combination ratio. If the first intersection-and-combination ratio is larger, it indicates that the area overlapped between "the area in the long jump frame where the long jumper is located" and "the long jump waiting area" is larger. On the contrary, if the first intersection-and-combination ratio is smaller, it indicates that the area overlapped between "the area in the long jump frame where the long jumper is located" and "the long jump waiting area" is smaller, so that it can be judged that the long jumper gradually moves away from the area near the take-off line.

[0061] Then, in step S123, the intersection and union ratio between the areas corresponding to the long jump personnel in each two adjacent long jump frames is obtained respectively. Here, "each two adjacent long jump frames" refers to each two adjacent long jump frames after the long jump frames are arranged in the order of time. "The intersection and union ratio between the areas corresponding to the long jump personnel in two adjacent long jump frames" refers to the intersection and union ratio between the two areas where the long jump personnel are located in two adjacent long jump frames.

[0062] For example, firstly, a plurality of long jump frames are arranged in chronological order, and the plurality of long jump frames include two adjacent long jump frames, specifically a long jump frame p1 and a long jump frame p2, and the long jump frame p1 and the long jump frame p2 are two frames of images adjacent in time. Next, the region a1 where the long jumper is located in the long jump frame p1 is obtained, and the region a2 where the long jumper is located in the long jump frame p2 is obtained. Then, the intersection-and-union ratio between the region a1 and the region a2 is determined, where the intersection-and-union ratio between the region a1 and the region a2 is the above-mentioned "intersection-and-union ratio between the regions corresponding to the long jumper in two adjacent long jump frames". For the convenience of description, the "intersection-and-union ratio between the regions corresponding to the long jumper in each two adjacent long jump frames" is referred to as the second intersection-and-union ratio.

[0063] In the embodiment disclosed herein, the second intersection-in-combination ratio can be used to measure the motion state of the long jumper. If the second intersection-in-combination ratio is larger, it indicates that the position of the area where the long jumper is located has changed greatly in two adjacent long jump frames. In other words, the movement speed of the long jumper is relatively large at this time, so it can be judged that the long jumper has not landed completely and smoothly. On the contrary, if the second intersection-in-combination ratio is smaller, it indicates that the position of the area where the long jumper is located has changed less in two adjacent long jump frames. In other words, the movement speed of the long jumper is relatively small at this time, so it can be judged that the long jumper has landed smoothly.

[0064] In the present embodiment, the intersection and union ratio between the area in the long jump frame and the long jump waiting area of ​​the long jump personnel when the long jump personnel is taking off or performing the preparatory action for taking off can be used as the "first threshold value". When the long jump personnel is taking off or performing the preparatory action for taking off, the long jump personnel will be located in the area near the take-off line, and the overlap between the area in the long jump frame and the long jump waiting area of ​​the long jump personnel is high. If the long jump personnel has taken off, that is, the long jump personnel has not contacted the long jump mat or the ground, the overlap between the area in the long jump frame and the long jump waiting area of ​​the long jump personnel is low.

[0065] It can be seen from the above description that when the first intersection-and-union ratio is greater than or equal to the first threshold value, it means that the overlapping area between "the area of ​​the long jumper in the long jump frame" and the "long jump waiting area" is very large, so it can be judged that the long jumper is very close to the area near the take-off line, that is, the long jumper is taking off or preparing for the take-off at this time, and the long jump frame does not belong to the take-off frame. When the first intersection-and-union ratio is less than the first threshold value, it means that the overlapping area between "the area of ​​the long jumper in the long jump frame" and the "long jump waiting area" is very small, so it can be judged that the long jumper is very far away from the area near the take-off line, that is, the long jumper is not taking off at this time, and the long jump frame does not belong to the take-off frame.

[0066] Therefore, when the first intersection-over-joint ratio is less than a preset first threshold, it can be determined that the corresponding long jump frame and / or the long jump frame before this time is the take-off frame, thereby implementing step S124.

[0067] In the embodiments of the present disclosure, the first threshold mentioned above can be set arbitrarily, and those skilled in the art can understand that the method of presetting the first threshold and the value of the first threshold are not subject to any limitation.

[0068] In other embodiments, it is also possible to determine whether a certain long jump frame is a take-off frame or a landing frame by determining the distance between a long jumper and a long jump pad or the ground. For example, when a long jumper is located at the take-off line annex and the distance between the long jumper and the long jump pad is zero, it means that the long jumper has not yet taken off, and the long jump frame where the long jumper is at this time is used as the take-off frame. Generally speaking, the first frame image in the video of the long jump, i.e., the first long jump frame is the take-off frame. When a long jumper is at a certain distance from the take-off line, and the distance between the long jumper and the long jump pad is zero, it means that the long jumper has landed, and the long jump frame where the long jumper is at this time is used as the landing frame.

[0069] Figure 3 An exemplary flowchart 130 of a method for identifying key points in a key frame and obtaining key point coordinates in some embodiments of the present disclosure is shown. It can be understood that the method for identifying key points in a key frame and obtaining key point coordinates is a specific implementation of the aforementioned step S130, so the aforementioned method is combined with Figure 1 and Figure 3 The features described can analogously apply here.

[0070] As shown in the figure, in step S131, the key points in the take-off frame are detected by the trained key point detection model to obtain the first key point, and the coordinates of the first key point are obtained; in step S132, the key points in the landing frame are detected by the key point detection model to obtain the second key point, and the coordinates of the second key point are obtained.

[0071] The key point detection model in step S131 can be a target detection model, for example, a pose estimation model (You Only Look Once v8 pose, yolov8-pose for short), which is used to detect key points of a human body. In step S131, the key point detection model is used to detect key points of a human body on the take-off frame, and the key point detected in the take-off frame is called a "first key point", and the position coordinates of the first key point in the take-off frame are called "first key point coordinates".

[0072] In some embodiments, the key points include at least one of the long jumper's left elbow, right elbow, left wrist, right wrist, left hip joint, right hip joint, left knee, right knee, left ankle, right ankle, left toe, right toe, left heel and right heel.

[0073] Similar to step S131, in step S132, the key points of the human body in the landing frame are detected by the above-mentioned key point detection model, and the key points detected in the landing frame are called "second key points", and the position coordinates of the second key points in the landing frame are called "second key point coordinates".

[0074] In some embodiments, the key point detection model in step S131 is trained in the following manner: first, one or more long jump sample videos are obtained, each long jump sample video includes one or more long jump sample frames; then, the key points in each long jump sample frame are marked to obtain a long jump marked sample frame; then, the initial key point detection model is trained by the long jump marked sample frame and the long jump sample frame to obtain the trained key point detection model.

[0075] In the process of training the key point detection model, "long jump sample video" refers to a video of any long jumper performing long jump shot in the aforementioned step S110. "Long jump sample frame" refers to a frame of image in the long jump sample video. "Long jump annotated sample frame" refers to the image after the key points of the human body are annotated in the long jump sample frame. "Initial key point detection model" refers to a key point detection model that has not been trained, that is, the key point detection model before training. The structure of the initial key point detection model is the same as that of the key point detection model, and the parameters of the initial key point detection model are different from those of the key point detection model.

[0076] In some embodiments, the above-mentioned training of the initial key point detection model by the long jump labeled sample frame and the long jump sample frame to obtain the trained key point detection model specifically includes: first, dividing the long jump sample frame and the long jump labeled sample frame into a verification set, a training set and a test set according to a preset ratio. The initial key point detection model is trained by the verification set, the training set and the test set, and the parameters of the initial key point detection model are adjusted. Finally, the trained initial key point detection model is used as the trained key point detection model.

[0077] Figure 4 An exemplary flowchart 140 of a method for determining whether a long jump is qualified in some embodiments of the present disclosure is shown. It can be understood that the method for determining whether a long jump is qualified is a specific implementation of the aforementioned step S140, so the aforementioned method is combined with Figure 1 and Figure 4 The features described can analogously apply here.

[0078] As shown in the figure, in step S141, the take-off frame is perspective transformed to obtain the orthographic projection take-off frame; in step S142, the first key point transformation coordinates in the orthographic projection take-off frame are determined according to the first key point coordinates, the first key point includes the left toe and the right toe, and the first key point transformation coordinates include the left toe transformation coordinates and the right toe coordinates; in step S143, according to the left toe transformation coordinates, the right toe transformation coordinates and the take-off line coordinates corresponding to the take-off line in the orthographic projection take-off frame, it is determined whether the left toe and the right toe do not exceed the take-off line; if the left toe and the right toe do not exceed the take-off line, then in step S144, it is determined that the take-off is qualified; otherwise, in step S145, it is determined that the take-off is unqualified.

[0079] "Perspective transformation" refers to projecting an image onto a new plane. In step S141, perspective transformation of the take-off frame refers to spatial coordinate transformation of the take-off frame to convert the take-off frame into an image equivalent to an image obtained by looking down from directly above the long jump mat. The image after perspective transformation of the take-off frame is called "orthographic projection take-off frame". For example, if the shape of the long jump mat in the take-off frame is a trapezoid, the spatial coordinate transformation of the take-off frame is performed so that the shape of the long jump mat is a rectangle. Those skilled in the art will understand that perspective transformation is achieved by matrix multiplication, and the specific process of perspective transformation is not subject to any limitation.

[0080] In some embodiments of the present disclosure, the shooting device is set on one of the two sides of the long jump mat, facing the side of a person performing the long jump, so the take-off frame is a side image of the long jump mat. Due to the perspective relationship, the shape of a rectangular long jump mat in the real world is a trapezoid or other non-rectangular shape in the take-off frame. Through step S141, the long jump mat whose shape is not rectangular in the take-off frame is transformed into a coordinate to obtain an image corresponding to the rectangular long jump mat, and the image corresponding to the rectangular long jump mat is referred to as the "orthographic projection take-off frame" here. The orthographic projection take-off frame is equivalent to the orthographic projection image taken from directly above the long jump mat.

[0081] The first key point coordinates in step S142 refer to the coordinates corresponding to the first key point in the take-off frame. The first key point also includes the left toe and the right toe of the long jumper. The coordinates of the left toe in the take-off frame are called "left toe coordinates", and the coordinates of the right toe in the take-off frame are called "right toe coordinates".

[0082] In step S141, perspective transformation is performed on the take-off frame, i.e., coordinate transformation is performed, and the coordinates of the first key point are also transformed in the same manner to obtain the corresponding coordinates of the first key point in the orthographic projection take-off frame. Then, after the first key point has undergone the above-mentioned coordinate transformation, step S142 is performed, i.e., the corresponding coordinates of the first key point in the orthographic projection take-off frame are determined, where the "corresponding coordinates of the first key point in the orthographic projection take-off frame" are referred to as "first key point transformation coordinates". The coordinates of the left toe in the orthographic projection take-off frame are referred to as "left toe transformation coordinates", and the coordinates of the right toe in the orthographic projection take-off frame are referred to as "right toe transformation coordinates".

[0083] Then, in step S143, by the take-off line coordinates corresponding to the take-off line in the take-off frame of the left toe transformation coordinates, the right toe transformation coordinates and the orthographic projection, the positional relationship between the left toe, the right toe and the take-off line in the orthographic projection take-off frame is judged, thereby determining whether the left toe and the right toe all exceed the take-off line in the long jump direction. If the left toe and the right toe do not exceed the take-off line, it means that the take-off point of the long jump personnel has not crossed the take-off line, and step S144 is executed at this time, i.e., it is determined that the take-off of the long jump personnel is qualified. If at least one of the left toe and the right toe exceeds the take-off line, it means that the take-off point of the long jump personnel has crossed the take-off line, and step S145 is executed at this time, i.e., it is determined that the take-off of the long jump personnel is unqualified.

[0084] For example, the direction of increasing distance along the scale line is called the positive direction of long jump, and a coordinate axis is established in the orthographic projection take-off frame. The established coordinate axis is called the "distance coordinate axis" here. The positive direction of the distance coordinate axis is the same as the positive direction of long jump, and the zero scale line of the distance coordinate axis is the position of the take-off line. The coordinate of the left toe on the distance coordinate axis is ld1, and the coordinate of the right toe on the distance coordinate axis is rd1. In the above case, if ld1 is less than zero and rd1 is less than zero, it means that in the long jump direction, neither the left toe nor the right toe exceeds the take-off line, so it can be judged that the long jumper's take-off is qualified; if ld1 is greater than zero and rd1 is less than zero, it means that in the long jump direction, the left toe exceeds the take-off line, so it can be judged that the long jumper's take-off is unqualified.

[0085] Figure 5 An exemplary flowchart 150 of a method for determining an initial long jump distance in some embodiments of the present disclosure is shown. It can be understood that the method for determining an initial long jump distance is a specific implementation of the aforementioned step S150, so the aforementioned method is combined with Figure 1 and Figure 5 The features described can analogously apply here.

[0086] As shown in the figure, in step S151, the landing frame is perspective transformed to obtain the orthographic projection landing frame; in step S152, the scale line transformation coordinates corresponding to the scale line coordinates in the orthographic projection landing frame are obtained; in step S153, according to the second key point coordinates, the first landing point in the orthographic projection landing frame and the first landing point coordinates corresponding to the first landing point are obtained; in step S154, the orthographic projection landing frame and the background frame are differentially operated to obtain a differential image; in step S155, the differential image is edge detected to extract the edge image of the long jumper in the differential image; In step S156, the key point in the edge image of the long jumper that is closest to the take-off line and lands on the ground is taken as the second landing point, and the second landing point coordinates corresponding to the second landing point are obtained; in response to the distance between the first landing point and the second landing point being greater than or equal to a preset third threshold, in step S157, the initial long jump distance is determined based on the first landing point coordinates and the scale line transformation coordinates; in response to the distance between the first landing point and the second landing point being less than the third threshold, in step S158, the initial long jump distance is determined based on the first landing point coordinates, the second landing point coordinates and the scale line transformation coordinates.

[0087] The method of perspective transformation of the landing frame in step S151 is the same as the method of perspective transformation of the take-off frame in step S141, which is not described here. The image obtained by perspective transformation of the landing frame in step S151 is called "orthographic projection landing frame".

[0088] When the perspective transformation is performed on the landing frame, the scale lines in the landing frame will be transformed in the same perspective transformation. The coordinates corresponding to the scale lines in the orthographic projection landing frame are referred to as "scale line transformation coordinates". Step S152 is executed to extract the scale line transformation coordinates in the orthographic projection landing frame.

[0089] Next, execute step S153 to determine the landing point in the second key point, that is, obtain the second key point in contact with the long jump mat or the ground in the orthographic projection landing frame. Here, the landing point in the second key point closest to the take-off point is called the "first landing point", and the coordinates of the first landing point in the orthographic projection landing frame are determined. The coordinates of the first landing point in the orthographic projection landing frame are called "second landing point coordinates".

[0090] The background frame in step S154 refers to the image corresponding to the background of the long jump. In some embodiments of the present disclosure, the long jump field without long jump personnel can be photographed by the aforementioned shooting equipment to obtain the background frame. The long jump field to be performed is a long jump mat or ground. It should be noted that the method of shooting the background frame is the same as the method of shooting the above-mentioned long jump frame. Specifically, the position of the shooting equipment when shooting the background frame is the same as the position of the shooting equipment when shooting the long jump frame, and the angle of shooting the background frame is the same as the angle of shooting the long jump frame. Therefore, the size and size of the background frame and the long jump frame are all the same.

[0091] In step S154, an image difference operation is performed on the forward projection landing frame and the background frame, that is, the pixel values ​​of the corresponding positions in the forward projection landing frame and the background frame are subtracted to weaken the similar parts of the image and highlight the changed parts of the image. The image obtained by the above-mentioned difference operation is called a "differential image". In step S155, the differential image is edge detected to extract the edge image of the long jumper in the differential image, that is, the outline of the long jumper in the differential image, and the size and size of the edge image, the background frame and the take-off frame are all the same. The edge detection here is used to extract the edge information of the differential image. It can be understood by those skilled in the art that the specific method of edge detection is not subject to any limitation.

[0092] In order to distinguish it from the first landing point mentioned above, the point in the edge image closest to the take-off line and in contact with the long jump mat or the ground is called the "second landing point", and the coordinates corresponding to the second landing point in the edge image are called "second landing point coordinates".

[0093] It can be seen from the description of the above embodiment that the first landing point and the second landing point are determined by two different methods, and the first landing point and the second landing point may overlap or may not overlap. In the case where the first landing point and the second landing point do not overlap, if the distance between the first landing point and the second landing point is too different, it means that at least one of the methods of determining the first landing point and determining the second landing point cannot accurately represent the landing point of the final effective score of the long jump. At this time, the second landing point is discarded and step S157 is executed, that is, the positional relationship between the first landing point and the scale line is determined by transforming the coordinates of the first landing point and the scale line; then, the initial long jump distance is determined by the positional relationship between the first landing point and the scale line.

[0094] The "third threshold" mentioned above refers to a critical value for judging whether the distance between the first landing point and the second landing point is too large. If the distance between the first landing point and the second landing point is greater than the third threshold, it indicates that the distance between the first landing point and the second landing point is too large. Otherwise, it indicates that the distance between the first landing point and the second landing point is small. Those skilled in the art can understand that the determination method and specific value of the third threshold can be set arbitrarily without any limitation.

[0095] If the distance between the first landing point and the second landing point is greater than or equal to the third threshold, it indicates that the distance between the first landing point and the second landing point is large, and step S157 is executed, that is, the positional relationship between the first landing point and the scale line is determined by transforming the coordinates of the first landing point and the scale line; then, the initial long jump distance is determined by the positional relationship between the first landing point and the scale line. For example, if the first landing point is located on the extension line of the scale line corresponding to the distance value of 1.5 meters, the initial long jump score of the long jumper is determined to be 1.5 meters.

[0096] If the distance between the first landing point and the second landing point is less than the third threshold, it indicates that the distance between the first landing point and the second landing point is small, and step S158 is executed at this time, that is, firstly obtaining the middle point between the first landing point and the second landing point, and the "middle point between the first landing point and the second landing point" is called the middle landing point. Then, the coordinates of the middle landing point in the orthographic projection landing frame are determined, and the "coordinates of the middle landing point in the orthographic projection landing frame" are called the middle landing point coordinates. Next, the initial long jump distance is determined by the relationship between the middle landing point and the scale line.

[0097] It should be noted that the specific method of determining the initial long jump distance through the relationship between the first landing point and the scale line is the same as the specific method of determining the initial long jump distance through the relationship between the middle landing point and the scale line, and will not be described in detail here. In the embodiment of the present disclosure, the middle landing point is used as the above-mentioned first landing point, and the specific method of determining the initial long jump distance through the relationship between the first landing point and the scale line is used to achieve the determination of the initial long jump distance through the relationship between the middle landing point and the scale line.

[0098] Figure 6 An exemplary flowchart 157 of a method for determining an initial long jump distance according to some embodiments of the present disclosure is shown. It can be understood that the method for determining the initial long jump distance according to the coordinates of the first landing point and the coordinates of the scale line is a specific implementation of the aforementioned step S157, so the aforementioned method is combined with Figure 1 , Figure 5 and Figure 6 The features described can be similarly applied here. The positional relationship between the first landing point and the scale lines includes that the first landing point is located on an extension line of a scale line and that the first landing point is not located on an extension line of any scale line.

[0099] As shown in the figure, in step S1571, in response to the first landing point being located on the extension line of a scale line, the initial long jump distance is determined by the distance value corresponding to the scale line; in step S1572, in response to the first landing point not falling on the extension line of any scale line, the image unit distance is determined by the spacing between adjacent scale lines and the distance corresponding to the spacing; in step S1573, the initial long jump distance is determined by the image unit distance and the length between the first landing point and the take-off line in the orthographic landing frame.

[0100] If the first landing point is located on an extension line of a scale line, then in step S1571, the initial long jump distance is determined by the distance value corresponding to the scale line. For example, if the first landing point is located on an extension line of a scale line corresponding to a distance value of 2.5 meters, the initial long jump score of the long jumper is determined to be 2.5 meters.

[0101] Furthermore, if the first landing point does not fall on the extension line of any scale line, then in step S1572, the image unit distance is determined by the spacing between adjacent scale lines and the distance corresponding to the spacing. In step S1573, the actual distance between the first landing point and the take-off line in the real space, that is, the initial long jump distance, is determined by the image unit distance and the length between the first landing point and the take-off line in the orthographic projection landing frame.

[0102] When the first landing point does not fall on the extension line of any scale line, it means that the actual distance between the first landing point and the take-off line in the real space, that is, the initial long jump distance, cannot be directly determined by the distance value corresponding to the scale line. For example, the first landing point is located between the extension line of the scale line corresponding to the distance value of 1.5 meters and the extension line of the scale line corresponding to the distance value of 2 meters. At this time, execute step S1573, and divide the distance values ​​corresponding to the above two scale lines by the spacing between any two adjacent scale lines in the orthographic projection landing frame to obtain the actual distance corresponding to each unit length in the orthographic projection landing frame in the real space. For the convenience of description, the "actual distance corresponding to each unit length in the orthographic projection landing frame in the real space" is referred to as the image unit distance.

[0103] Then, in step S1573, the image unit distance and the length between the first landing point and the take-off line in the orthographic projection landing frame are multiplied to determine the actual distance between the first landing point and the take-off line in the real space, that is, the initial long jump distance.

[0104] For example, if in the orthographic projection landing frame, the length between the first landing point and the take-off line in the orthographic projection landing frame is 17 pixels, and the length between the scale line corresponding to the distance value of 1.5 meters and the scale line corresponding to the distance value of 2 meters is 5 pixels, then the actual distance in the real space represented by each unit pixel is (2-1.5) / 5=0.1 meters, that is, the image unit distance is 0.1 meters. Then, through the image unit distance and the length between the first landing point and the take-off line in the orthographic projection landing frame, it is determined that the actual distance between the first landing point and the take-off line in the real space is 0.1×17=1.7 meters, so the initial long jump distance in this embodiment is 1.7 meters.

[0105] Figure 7 An exemplary flowchart 158 ​​of a method for determining an initial long jump distance according to other embodiments of the present disclosure is shown. It can be understood that the method for determining the initial long jump distance according to the first landing point coordinates, the second landing point coordinates and the scale line coordinates is a specific implementation of the aforementioned step S158, so the aforementioned method is combined with Figure 1 , Figure 5 and Figure 7 The features described can analogously apply here.

[0106] As shown in the figure, in step S1581, the middle point between the first landing point and the second landing point is determined according to the first landing point coordinates and the second landing point coordinates, and the middle point coordinates corresponding to the middle point are obtained; in response to the middle point being located on the extension line of a scale line, in step S1582, the initial long jump distance is determined by the distance value corresponding to the scale line; if the middle point does not fall on the extension line of any scale line, in step S1583, the image unit distance is determined by the spacing between adjacent scale lines and the distance corresponding to the spacing; in step S1584, the initial long jump distance is determined by the image unit distance and the length between the middle point of the orthographic landing frame and the take-off line.

[0107] The middle point in step S1581 is the midpoint between the first landing point and the second landing point in the orthographic projection landing frame, and the distance between the middle point and the first landing point is equal to the distance between the middle point and the second landing point. In the orthographic projection landing frame, the coordinates corresponding to the middle point are "middle point coordinates".

[0108] The process of step S1582 is the same as the process of step S1571 mentioned above, and the process of step S1583 is the same as the process of step S1572 mentioned above, and will not be repeated here.

[0109] The process of step S1584 specifically includes: multiplying the image unit distance and the length between the middle point of the orthographic projection landing frame and the take-off line to obtain the actual distance between the middle point and the take-off line in the real space, that is, the initial long jump distance.

[0110] Figure 8 An exemplary flowchart 160 of a method for performing error correction on an initial long jump distance according to some embodiments of the present disclosure is shown. It can be understood that the method for performing error correction on an initial long jump distance is a specific implementation of the aforementioned step S160, so the aforementioned method is combined with Figure 1 and Figure 8 The features described can analogously apply here.

[0111] As shown in the figure, in step S161, the mapping relationship between the initial long jump distance and the measurement error is determined; in step S162, the measurement error corresponding to the initial long jump distance is determined according to the mapping relationship; in step S163, the initial long jump distance is error corrected according to the measurement error to obtain the long jump distance.

[0112] In some embodiments, in step S161, the determining of the mapping relationship between the initial long jump distance and the measurement error includes: setting multiple groups of long jump video samples and the actual long jump distance of the long jumper in each group of long jump video samples; performing long jump detection on the multiple groups of long jump video samples to obtain predicted distance samples; performing subtraction operation based on the actual long jump distance and the predicted distance samples to determine the measurement sample error corresponding to each group of long jump video samples; using the predicted distance samples and the measurement sample errors as training sets, and training a preliminary mapping determination model based on the training set to obtain a mapping determination model; determining the mapping relationship between the initial long jump distance and the measurement error through the mapping determination model.

[0113] In this embodiment, the "measurement error" refers to the difference between the initial long jump distance obtained by the above steps and the actual long jump distance of the long jumper. Here, the "corresponding relationship between the initial long jump distance and the measurement error" is referred to as the mapping relationship between the initial long jump distance and the measurement error. For example, the mapping relationship between the initial long jump distance and the measurement error is defined as f(.), the initial long jump distance is defined as jl, and the measurement error e is defined, then e=f(jl).

[0114] In some embodiments, the specific process of step S161 includes: first, setting multiple groups of long jump video samples and the actual long jump distance of the long jumper in each group of long jump video samples, where the "long jump video sample" refers to a video of any long jumper performing a long jump, and the "actual long jump distance" refers to the actual long jump distance of the long jumper in the real three-dimensional space in each group of long jump video samples.

[0115] Secondly, long jump detection is performed on multiple groups of long jump video samples using the aforementioned method for detecting long jump to obtain predicted distance samples, here "predicted distance samples".

[0116] Next, a subtraction operation is performed based on the actual long jump distance and the predicted distance samples to determine the measurement sample error corresponding to each group of long jump video samples. Here, the “measurement sample error” refers to the difference between the actual long jump distance and the predicted distance samples.

[0117] Then, the predicted distance sample and the measurement sample error are used as a training set, and the preliminary mapping determination model is trained according to the training set to obtain the mapping determination model. Specifically, the predicted distance sample is used as the input of the preliminary mapping determination model, and the measurement sample error is used as the input of the preliminary mapping determination model, and the preliminary mapping determination model is trained to obtain the mapping determination model. The preliminary mapping determination model can be a regression model in the field of machine learning, for example, a polynomial regression model. The mapping determination model can be a regression model in the field of machine learning, for example, a polynomial regression model. The mapping determination model is a trained preliminary mapping determination model. The mapping determination model fits the mapping relationship between the initial long jump distance and the measurement error, and the mapping relationship between the initial long jump distance and the measurement error can be obtained through the mapping determination model.

[0118] In step S162, the measurement error corresponding to the current initial long jump distance is determined by the initial long jump distance obtained in step S150 and the mapping relationship between the initial long jump distance and the measurement error determined above. For example, if the mapping relationship between the initial long jump distance and the measurement error is that the measurement error is 1% of the initial long jump distance, then the measurement error corresponding to the current initial long jump distance is 1% of the value of the current initial long jump distance.

[0119] Since the measurement error corresponding to the initial long jump distance has been obtained through steps S161-S162, and the measurement error is obtained by subtracting the initial long jump distance from the actual long jump distance, in step S163, the initial long jump distance can be corrected according to the measurement error, that is, the initial long jump distance and the measurement error are added, and the sum of the addition is called the "long jump distance". The long jump distance is the final score of the long jumper.

[0120] In summary, the method for detecting long jump provided by the disclosed embodiment first extracts key frames in the long jump video; then, identifies key points in the key frames and extracts key coordinates of the key points; then, according to the scale line coordinates and key point coordinates of the long jump mat in the key frames, it is possible to determine whether the long jump is qualified. And, in the case of qualified long jump, the initial long jump distance is determined according to the scale line coordinates and key point coordinates; finally, the initial long jump distance is error corrected to obtain the long jump distance, which can reduce the error of measuring the long jump distance, thereby improving the accuracy of measuring the long jump. Moreover, the above-mentioned method for measuring long jump can reduce manual participation, thereby simplifying the process of measuring long jump, so as to improve the efficiency of measuring long jump.

[0121] The disclosed embodiment also provides a device for detecting long jump. Fig. 9 An exemplary structural block diagram of an apparatus 900 for detecting long jump according to an embodiment of the present disclosure is shown.

[0122] As shown in the figure, the apparatus 900 for detecting long jump includes: a processor 910 and a memory 920 .

[0123] In the embodiment of the present disclosure, the processor 910 is configured to execute program instructions. The memory 920 is configured to store program instructions, and when the program instructions are loaded and executed by the processor 910, the apparatus 900 for detecting long jumps executes the aforementioned method for detecting long jumps.

[0124] In summary, the device 900 for detecting long jump provided by the embodiment of the present disclosure, when the program instruction is loaded by the processor 910 and executes the aforementioned method for detecting long jump, the device 900 for detecting long jump can determine whether the long jump is qualified by the scale line coordinates and key point coordinates of the long jump mat in the key frame. Moreover, when the long jump is qualified, the initial long jump distance is determined according to the scale line coordinates and the key point coordinates, and the initial long jump distance is error corrected to obtain the long jump distance, so that the error of measuring the long jump distance can be reduced, thereby improving the accuracy of measuring the long jump. At the same time, the method can reduce manual participation, thereby simplifying the process of measuring the long jump, to improve the efficiency of measuring the long jump.

[0125] The disclosed embodiment further provides a computer-readable medium having a computer program stored thereon, and when the computer program is executed by a processor, the processor executes the method for detecting long jump provided in the first aspect.

[0126] Although multiple embodiments of the present disclosure have been shown and described herein, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art may think of many changes, modifications, and alternatives without departing from the thought and spirit of the present disclosure. It should be understood that in the process of practicing the present disclosure, various alternatives to the embodiments of the present disclosure described herein may be adopted. The attached claims are intended to define the scope of protection of the present disclosure, and therefore cover equivalents or alternatives within the scope of these claims.

Claims

1. A method for detecting long jump, characterized in that: include: Get the video of the long jump; Extract key frames from the video, where the key frames include a take-off frame and a landing frame; Identify a first key point in the take-off frame and a second key point in the landing frame, and obtain first key point coordinates corresponding to the first key point and second key point coordinates corresponding to the second key point; Determining whether the long jump is qualified according to the scale line coordinates of the long jump mat in the key frame and the coordinates of the first key point; If the long jump is qualified, determining an initial long jump distance according to the scale line coordinates and the second key point coordinates; The initial long jump distance is error corrected to obtain the long jump distance.

2. The method according to claim 1, characterized in that The video includes one or more long jump frames, and extracting key frames from the video includes: A long jump waiting area is set in the long jump frame, and the long jump waiting area is located in the long jump waiting area of ​​the long jump mat; Determine an intersection-and-union ratio of the long jumper and the long jump waiting area in the long jump frame to obtain a first intersection-and-union ratio; and determining an intersection-and-union ratio between every two adjacent long jump frames to obtain a second intersection-and-union ratio; In response to the first intersection-and-joint ratio being less than a preset first threshold, taking the long jump frame corresponding to the first intersection-and-joint ratio as the take-off frame; In response to the second intersection-and-joint ratio being greater than a preset second threshold, the long jump frame corresponding to the second intersection-and-joint ratio is used as the landing frame.

3. The method according to claim 1, characterized in that The video includes one or more long jump frames, and the long jump frames include one or more key points, and wherein the identifying a first key point in the take-off frame and a second key point in the landing frame, and obtaining a first key point coordinate corresponding to the first key point and a second key point coordinate corresponding to the second key point includes: Detecting the key point in the take-off frame by using a trained key point detection model to obtain a first key point, and acquiring the coordinates of the first key point; and Detecting the key point in the landing frame by using the key point detection model to obtain a second key point, and acquiring the coordinates of the second key point; The key point detection model is trained in the following way: Acquire one or more long jump sample videos, each of the long jump sample videos including one or more long jump sample frames; Marking the key points in each of the long jump sample frames to obtain a long jump marked sample frame; Training the initial key point detection model through the long jump labeled sample frame and the long jump sample frame to obtain the trained key point detection model; Among them, the key points include at least one of the long jumper's left elbow, right elbow, left wrist, right wrist, left hip joint, right hip joint, left knee, right knee, left ankle, right ankle, left toe, right toe, left heel and right heel.

4. The method according to claim 1, characterized in that: Determining whether the long jump is qualified according to the scale line coordinates of the long jump mat in the key frame and the first key point coordinates includes: Performing perspective transformation on the take-off frame to obtain an orthographic projection take-off frame; Determine, according to the first key point coordinates, the first key point transformation coordinates in the orthographic projection take-off frame, wherein the first key point includes a left toe and a right toe, and the first key point transformation coordinates include a left toe transformation coordinate and a right toe coordinate; Determine whether both the left toe and the right toe do not exceed the take-off line according to the transformed coordinates of the left toe, the transformed coordinates of the right toe and the take-off line coordinates corresponding to the take-off line in the orthographic projection take-off frame; If the left toe and the right toe do not exceed the take-off line, the take-off is determined to be qualified; Otherwise, the take-off is determined to be unqualified.

5. The method according to claim 1, characterized in that: The background frame is an image corresponding to the background of the long jump, and when the long jump is qualified, determining the initial long jump distance according to the scale line coordinates and the second key point coordinates includes: Performing perspective transformation on the landing frame to obtain an orthographic projection landing frame; Obtaining the scale line transformation coordinates corresponding to the scale line coordinates in the orthographic projection landing frame; According to the coordinates of the second key point, obtaining the first landing point in the orthographic projection landing frame and the first landing point coordinates corresponding to the first landing point; Performing a difference operation on the orthographic projection landing frame and the background frame to obtain a difference image; Performing edge detection on the differential image to extract an edge image of the long jumper in the differential image; The key point in the edge image of the long jumper that is closest to the take-off line and lands on the ground is taken as the second landing point, and the coordinates of the second landing point corresponding to the second landing point are obtained; In response to the distance between the first landing point and the second landing point being greater than or equal to a preset third threshold, determining the initial long jump distance according to the coordinates of the first landing point and the scale line transformation coordinates; In response to the distance between the first landing point and the second landing point being less than the third threshold, the initial long jump distance is determined according to the first landing point coordinates, the second landing point coordinates and the scale line transformation coordinates.

6. The method according to claim 5, characterized in that Determining the initial long jump distance according to the first landing point coordinates and the scale line coordinates comprises: In response to the first landing point being located on an extension line of the scale line, an initial long jump distance is determined by passing a distance value corresponding to the scale line; In response to the first landing point not falling on an extension line of any of the scale lines, determining an image unit distance by using a spacing between adjacent scale lines and a distance corresponding to the spacing; An initial long jump distance is determined by the image unit distance and the length between the first landing point and the take-off line in the orthographic projection landing frame.

7. The method according to claim 5, characterized in that Determining the initial long jump distance according to the first landing point coordinates, the second landing point coordinates and the scale line coordinates comprises: Determine a middle point between the first landing point and the second landing point according to the first landing point coordinates and the second landing point coordinates, and obtain middle point coordinates corresponding to the middle point; In response to the middle point being located on an extension line of one of the scale lines, determining an initial long jump distance by a distance value corresponding to the scale line; In response to the middle point not falling on an extension line of any scale line, determining the image unit distance by a spacing between adjacent scale lines and a distance corresponding to the spacing; The initial long jump distance is determined by the image unit distance and the length between the middle point of the orthographic projection landing frame and the take-off line.

8. The method according to claim 1, characterized in that The error correction of the initial long jump distance to obtain the long jump distance comprises: Determining a mapping relationship between the initial long jump distance and the measurement error; Determine the measurement error corresponding to the initial long jump distance according to the mapping relationship; Performing error correction on the initial long jump distance according to the measurement error to obtain the long jump distance; Wherein, determining the mapping relationship between the initial long jump distance and the measurement error comprises: Setting a plurality of groups of long jump video samples and the actual long jump distance of the long jumper in each group of the long jump video samples; Performing long jump detection on a plurality of groups of long jump video samples to obtain predicted distance samples; Performing a subtraction operation based on the actual long jump distance and the predicted distance samples to determine the measurement sample error corresponding to each group of the long jump video samples; The predicted distance samples and the measured sample errors are used as training sets, and a preliminary mapping determination model is trained according to the training sets to obtain a mapping determination model; The mapping relationship between the initial long jump distance and the measurement error is determined by the mapping determination model.

9. A device for detecting long jump, characterized in that: include: a processor configured to execute program instructions; as well as A memory configured to store the program instructions, which, when loaded and executed by the processor, causes the apparatus to perform the method according to any one of claims 1 to 8.

10. A computer-readable medium, characterized in that The computer readable medium stores a computer program, and when the computer program is executed by a processor, the processor executes the method according to any one of claims 1 to 8.

Images