Standing long jump distance measuring method and device, terminal and storage medium

Abstract

The embodiment of the invention provides a standing long jump distance measuring method and device, a terminal and a storage medium. The standing long jump distance measurement method comprises the steps of obtaining a first video frame; inputting the first video frame into a first classification model to obtain a human body frame confidence coefficient of the at least one moving object and an intersection-to-union ratio between the human body frame and a long jump pad target frame; the intersection-union ratio is a ratio of an intersection area to a union area, the intersection area is an overlapped area between the human body frame and the long jump pad target frame, and the union area is a union area between the human body frame and the long jump pad target frame; under the condition that the intersection-to-union ratio is greater than or equal to a first threshold value and the human body frame confidence coefficient is greater than or equal to a second threshold value, determining a motion area list of the motion object; one motion area list comprises a motion area of at least one motion object; the long jump distance of the motion object is determined based on the motion area list; therefore, the standing long jump test can be carried out by multiple persons at the same time, and the hardware cost is reduced.

Description

Technical Field

[0001] The present invention relates to, but is not limited to, the field of artificial intelligence technology, and particularly relates to a standing long jump distance measurement method and device, a terminal, and a storage medium. Background Art

[0002] Currently, distance measurement methods are mainly divided into three categories: manual, hardware sensing devices, and artificial intelligence.

[0003] Among them, for the manual distance measurement method, each person's subjective standard is different, and it is highly dependent on the visual experience of the coach; moreover, when multiple testers perform standing long jump movements together, it is very labor-intensive and the evaluation efficiency is low.

[0004] The distance measurement method based on hardware sensing devices often requires a set of devices for each person, which is very bulky and the device cost is high; and testers need to wear certain hardware devices, which affects the movement effect; in addition, for different testers, conditions such as height and / or body fat are inconsistent, and the long jump results are easily interfered.

[0005] For the standing long jump distance measurement method based on artificial intelligence, it cannot solve the problem of insufficient ranging accuracy caused by the jitter of the human key points of the tester; nor can it solve problems such as multiple people performing standing long jump movements, easy interference, and difficulty in identifying illegal actions. Summary of the Invention

[0006] In view of this, embodiments of the present invention provide a standing long jump distance measurement method and device, a terminal, and a storage medium to solve the above technical problems.

[0007] The technical solution of the present invention is implemented as follows:

[0008] In a first aspect, an embodiment of the present invention provides a standing long jump distance measurement method, including:

[0009] Obtain a first video frame;

[0010] Input the first video frame into a first classification model to obtain the confidence of the human body frame of at least one moving object and the intersection over union (IoU) between the human body frame and the target frame of the long jump mat; wherein, the IoU is the ratio of the intersection area to the union area, the intersection area is the overlapping area between the human body frame and the target frame of the long jump mat, and the union area is the union area between the human body frame and the target frame of the long jump mat;

[0011] When the IoU is greater than or equal to a first threshold and the confidence of the human body frame is greater than or equal to a second threshold, determine a list of movement areas of the moving object; wherein, a list of movement areas includes the movement areas of at least one moving object;

[0012] Based on the list of movement areas, determine the long jump distance of the moving object.

[0013] In the above solution, determining the long jump distance of the moving object based on the list of movement areas includes:

[0014] Obtaining at least one second video frame of the moving object long jumping in the movement area in the list of movement areas;

[0015] Inputting the second video frame into a second classification model to obtain the landing key frames of at least one of the moving objects;

[0016] Determining the long jump distance of the moving object based on the heel information in the first key point information of the landing key frame and the scale line information of the long jump mat.

[0017] In the above solution, the method further includes:

[0018] When the intersection over union is less than the first threshold and / or the confidence of the human body box is less than the second threshold, outputting a first prompt message; wherein, the first prompt message is used to prompt the moving object to move to or leave the movement area.

[0019] In the above solution, inputting the second video frame into the second classification model to obtain the landing key frames of at least one of the moving objects includes: inputting the second video frame into the second classification model to obtain the takeoff key frames and landing key frames of at least one of the moving objects;

[0020] The method further includes: when the toe tip information in the second key point information of the takeoff key frame exceeds the takeoff line, outputting a second prompt message; and / or when the hip information in the first key point information meets a predetermined condition, outputting the second prompt message; wherein, the second prompt message is used to indicate that the long jump is a violation action.

[0021] In the above solution, determining the long jump distance of the moving object based on the heel information in the first key point information of the landing key frame and the scale line information of the long jump mat includes:

[0022] When the toe tip information in the second key point information does not exceed the takeoff line and the hip information in the first key information does not meet the predetermined condition, determining the long jump distance of the moving object according to the heel information in the first key information and the scale information of the long jump mat.

[0023] In the above solution, inputting the second video frame into the second classification model to obtain the takeoff key frames and landing key frames of at least one of the moving objects includes:

[0024] Input the second video frame into the second classification model to obtain the first probability of the takeoff category and the second probability of the landing category for at least one video frame;

[0025] If the first probability of the video frame is greater than or equal to the second probability, determine that the video frame is a takeoff candidate frame; or, if the first probability of the video frame is less than the second probability, determine that the video frame is a landing candidate frame;

[0026] Based on at least one of the takeoff candidate frames, determine the takeoff key frame;

[0027] Based on at least one of the landing candidate frames, determine the landing key frame.

[0028] In the above solution, the determining the takeoff key frame based on at least one of the takeoff candidate frames includes one of the following:

[0029] If the difference between the maximum value and the minimum value of the first probability in at least two consecutive takeoff candidate frames is less than or equal to a third threshold, determine that the takeoff candidate frame corresponding to the intermediate value of the first probabilities of at least two takeoff candidate frames is the takeoff key frame;

[0030] If there is no difference between the first probabilities of two consecutive takeoff candidate frames that is less than or equal to the third threshold, determine that the takeoff candidate frame corresponding to the maximum value of the first probability is the takeoff key frame;

[0031] If the first probability of the takeoff candidate frame is greater than or equal to a fourth threshold, determine that the takeoff candidate frame is the takeoff key frame.

[0032] In the above solution, the determining the landing key frame based on at least one of the landing candidate frames includes one of the following:

[0033] If the difference between the maximum value and the minimum value of the second probability in at least two consecutive landing candidate frames is less than or equal to a fifth threshold, determine that the landing candidate frame corresponding to the intermediate value of the second probabilities of at least two landing candidate frames is the candidate key frame;

[0034] If there is no difference between the second probabilities of two consecutive landing candidate frames that is less than or equal to the fifth threshold, determine that the landing candidate frame corresponding to the maximum value of the second probability is the candidate key frame;

[0035] If the second probability of the landing candidate frame is greater than or equal to a sixth threshold, determine that the landing candidate frame is the candidate key frame.

[0036] In the above solution, determining the long jump distance of the moving object based on the heel information and the long jump mat scale line information in the first key point information of the landing key frame includes:

[0037] Based on the long jump mat scale line information, determine the first scale coordinate and the second scale coordinate that are closest to the heel information; the distance of the first scale coordinate from the takeoff line is closer than the distance of the second scale coordinate from the takeoff line;

[0038] Determine the first actual coordinate and the second actual coordinate corresponding to the first scale coordinate and the second scale coordinate on the long jump mat respectively;

[0039] Based on the first scale coordinate, the second scale coordinate, the first actual coordinate, and the second actual coordinate, determine the long jump distance.

[0040] In a second aspect, an apparatus for measuring the standing long jump distance according to an embodiment of the present invention includes:

[0041] An acquisition module, configured to acquire a first video frame;

[0042] A processing module, configured to input the first video frame into a first classification model to obtain the confidence of the human body frame of at least one moving object and the intersection over union (IoU) between the human body frame and the long jump mat target frame; wherein, the IoU is the ratio of the intersection area to the union area, the intersection area is the overlapping area between the human body frame and the long jump mat target frame, and the union area is the union area between the human body frame and the long jump mat target frame;

[0043] A determination module, configured to determine a list of movement areas of the moving object when the IoU is greater than or equal to a first threshold and the confidence of the human body frame is greater than or equal to a second threshold; wherein, one list of movement areas includes the movement areas of at least one of the moving objects;

[0044] The processing module is configured to determine the long jump distance of the moving object based on the list of movement areas.

[0045] In a third aspect, an embodiment of the present invention provides a terminal, which includes a processor and a memory for storing a computer program that can run on the processor; wherein, when the processor runs the computer program, the standing long jump distance measurement method according to any embodiment of the present invention is implemented.

[0046] In a fourth aspect, an embodiment of the present invention further provides a computer storage medium, in which there are computer-executable instructions, and the computer-executable instructions are executed by a processor to implement the standing long jump distance measurement method according to any embodiment of the present invention.

[0047] Fifth aspect, an embodiment of the present invention provides a computer program product, which includes a computer program or instruction. When the computer program or instruction is executed by a processor, it realizes any embodiment of the present invention 0

[0048] In an embodiment of the present invention, by obtaining a first video frame; inputting the first video frame into a first classification model to obtain the confidence of the human body frame of at least one moving object and the intersection over union (IoU) between the human body frame and the target frame of the long jump mat; wherein, the IoU is the ratio of the intersection area to the union area, the intersection area is the overlapping area between the human body frame and the target frame of the long jump mat, and the union area is the union area between the human body frame and the target frame of the long jump mat; in the case where the IoU is greater than or equal to a first threshold and the confidence of the human body frame is greater than or equal to a second threshold, determining a list of movement areas of the moving object; wherein, one list of movement areas includes the movement areas of at least one of the moving objects; based on the list of movement areas, determining the long jump distance of the moving object.

[0049] In this way, the embodiment of the present invention can determine the list of movement areas of the moving object without fixing the movement area of the moving object in advance, so as to determine the movement areas where multiple people perform standing long jump simultaneously, and realize the test of multiple people performing standing long jump simultaneously where multiple people can start moving at any position within the camera acquisition range of the terminal. Moreover, it is also possible to eliminate the need to wear any hardware devices, reducing the hardware cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic flowchart of the first standing long jump distance measurement method provided by the embodiment of the present invention.

[0051] Figure 2 It is a schematic diagram of collecting the first video frame from different orientations provided by the embodiment of the present invention.

[0052] Figure 3 It is a schematic flowchart of the second standing long jump distance measurement method provided by the embodiment of the present invention.

[0053] Figure 4 It is a schematic flowchart of the third standing long jump distance measurement method provided by the embodiment of the present invention.

[0054] Figure 5 It is a schematic diagram of a takeoff candidate frame and a landing candidate frame provided by the embodiment of the present invention.

[0055] Figure 6 It is a schematic flowchart of the fourth standing long jump distance measurement method provided by the embodiment of the present invention.

[0056] Figure 7 It is a schematic diagram of the standing long jump distance measurement calculation provided by the embodiment of the present invention.

[0057] Figure 8 This is a schematic flowchart of the fifth standing long jump distance measurement method provided by an embodiment of the present invention.

[0058] Figure 9 This is a schematic structural diagram of a standing long jump distance measurement device provided by an embodiment of the present invention.

[0059] Figure 10 This is a schematic hardware structure diagram of a terminal provided by an embodiment of the present invention. Detailed implementation manners

[0060] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0061] In subsequent descriptions, suffixes such as "module", "component", or "unit" used to represent elements are only for the convenience of explaining the present invention, and they have no specific meaning in themselves. Therefore, "module", "component", or "unit" can be used interchangeably. In addition, in subsequent descriptions, prefixes such as "first" or "second" used to identify information are only for the convenience of explaining the present invention, and they have no specific meaning in themselves. In addition, in subsequent descriptions, "a plurality of" means two or more; "a variety of" means two or more.

[0062] As Figure 1 shown, an embodiment of the present invention provides a standing long jump distance measurement method, including the following steps:

[0063] Step S11: Obtain a first video frame;

[0064] Step S12: Input the first video frame into a first classification model to obtain the confidence of the human body box of at least one moving object and the intersection over union (IoU) between the human body box and the target box of the long jump mat; where the IoU is the ratio of the intersection area to the union area, the intersection area is the overlapping area between the human body box and the target box of the long jump mat, and the union area is the union area between the human body box and the target box of the long jump mat;

[0065] Step S13: When the IoU is greater than or equal to a first threshold and the confidence of the human body box is greater than or equal to a second threshold, determine a list of movement areas of the moving object; where a list of movement areas includes the movement areas of at least one moving object;

[0066] Step S14: Determine the long jump distance of the moving object based on the list of movement areas.

[0067] The standing long jump distance measurement method provided by the embodiments of the present invention can be executed by a terminal; the terminal can be any type of mobile terminal or fixed terminal. For example, the terminal can be, but is not limited to, at least one of the following: a mobile communication device, a computer, a server, a tablet computer, a gaming device, a smart office device, an industrial device, and / or a wearable device, etc.

[0068] In some embodiments, step S11 includes: collecting a first video frame. Here, the terminal includes a camera module; the terminal can use the camera module to collect a first video frame including at least one moving object. The first video frame includes at least one video frame, and there is an image of at least one moving object in one video frame. Here, one video frame can be an image. In the embodiments of the present invention, at least one includes one or more; multiple means two or more.

[0069] In some embodiments, the first classification model can be any artificial intelligence model that can identify the target frame of the long jump mat, the human body frame, and the confidence level. For example, the first classification model can be a target detection model; the target detection model can be a YOLOX model, etc. Here, the target frame of the long jump mat can be the area where the long jump mat is located; the human body frame can be the area where the moving object is located. Here, the confidence level of the human body frame refers to the probability that there is a human body or a moving object in the human body frame. The relatively larger the confidence level of the human body frame, the more likely there is a human body or a moving object in the human body frame; the relatively smaller the confidence level of the human body frame, the less likely there is a human body or a moving object.

[0070] It can be understood that the intersection over union is usually the ratio of the intersection to the union. In the present invention, the intersection over union is the ratio of the intersection area to the union area; the intersection area refers to the intersection of the two areas of the human body frame and the target frame of the long jump mat; the union area refers to the union of the two areas of the human body frame and the target frame of the long jump mat.

[0071] In some embodiments, when the value range of the intersection over union is greater than or equal to 0 and less than or equal to 1, the first threshold can be a value greater than or equal to 0.7. For example, the first threshold is 0.78 or 0.8 or 0.85, etc.

[0072] In some embodiments, when the value range of the confidence level of the human body frame is greater than or equal to 0 and less than or equal to 1, the second threshold can be a value greater than or equal to 0.5. For example, the first threshold can be 0.5 or 0.6 or 0.7, etc.

[0073] In some embodiments, when determining the list of motion regions of the motion object in step S13, it includes: determining the list of motion regions of at least one motion object. For example, if the terminal determines that the intersection-over-union ratio of one motion object is greater than or equal to the first threshold and the confidence of the human body box is greater than or equal to the second threshold, it determines the list of motion regions of this one motion object; the list of motion regions includes one motion region of this motion object. Another example is that if the terminal determines that the intersection-over-union ratios of multiple motion objects are greater than or equal to the first threshold and the confidence of the human body box is greater than or equal to the second threshold, it determines the list of motion regions of these multiple motion objects; the list of motion regions includes the motion regions corresponding to the multiple motion objects respectively.

[0074] In some embodiments, when collecting the first video frame in step S11, the first video frame of at least one motion object can be collected from the front, or from the side, or at any angle. Here, collecting from the front, from the side, and at any angle respectively refer to collecting from the front, from the side, and at any angle with respect to at least one motion object or facing the length direction of the long jump mat target frame. If there are multiple motion objects, when collecting the first video frame in the side collection method, these multiple motion objects may overlap, then it is determined that the first video is not used to input the first classification model to determine the intersection-over-union ratio of the motion object or the intersection-over-union ratio of the overlapping objects is not determined.

[0075] Exemplarily, as Figure 2 shown, the first video frames of motion object 1, motion object 2, motion object 3, and motion object 4 are collected from the front; among them, the human body boxes corresponding to the motion object 1, motion object 2, motion object 3, and motion object 4 are human body box 1, human body box 2, human body box 3, and human body box 4 respectively; after the terminal inputs the first video frame into the first classification model, the intersection-over-union ratios between the human body boxes of the 4 motion objects and the long jump mat box can be determined respectively.

[0076] Based on the above embodiments, if the first video frames of the 4 motion objects are collected from the side, then the human body boxes 1, 3, and 4 corresponding to motion object 1, motion object 3, and motion object 4 respectively overlap with each other, then the intersection-over-union ratios of the 4 motion objects are not considered based on the first video, or only the intersection-over-union ratio of motion object 2 is determined.

[0077] In the embodiments of the present invention, the terminal can directly obtain the first video frame of the moving object, and obtain the moving area list of the moving object based on the first video frame. In this way, without fixing the moving area of the moving object in advance, the moving area list of the moving object can be determined, so that the moving areas where multiple people perform standing long jumps simultaneously can be determined, and the test of multiple people performing standing long jumps simultaneously can be realized where multiple people can start moving at any position within the camera acquisition range of the terminal. Moreover, it is not necessary to wear any hardware devices, reducing the hardware cost.

[0078] As Figure 3 shown, in some embodiments, step S14 includes:

[0079] Step S141: Obtain the second video frame of at least one moving object jumping in the moving area in the moving area list;

[0080] Step S142: Input the second video frame into the second classification model to obtain the landing key frames of at least one moving object;

[0081] Step S143: Determine the long jump distance of the moving object based on the heel information and the long jump mat scale line information in the first key point information of the landing key frame.

[0082] In some embodiments, step S141 may include: collecting the second video frame. Here, the terminal can collect the second video frame including at least one moving object by using the camera module. The second video frame includes at least one video frame, and there is an image of at least one moving object jumping in the moving area in one video frame. Optionally, the second video frame includes: takeoff candidate frame, landing candidate frame, takeoff key frame, and / or landing key frame, etc.

[0083] Optionally, the first video frame is different from the second video frame. The first video frame is the video frame collected based on the stage when the moving object is preparing to perform a long jump; the second video frame is the video frame collected when the moving object is performing a long jump.

[0084] Here, the takeoff candidate frame and the takeoff key frame are video frames for describing the takeoff posture of the moving object; the landing candidate frame and the landing key frame are video frames for describing the landing posture of the moving object.

[0085] In some embodiments, the second classification model can be any key posture recognition model. The key posture includes the takeoff and / or landing key postures. The second classification model can be, for example, a VGG classification model, etc.

[0086] In some embodiments, step S142 may include: inputting the second video frame into a second classification model to determine at least one landing candidate frame of the moving object; and determining a landing key frame of the moving object based on the at least one landing candidate frame. Here, one moving object corresponds to at least one landing candidate frame; in step S142, at least one landing candidate frame corresponding to one or more moving objects may be determined, and a landing key frame corresponding to one or more moving objects may be determined. Here, when there are multiple moving objects currently, the landing candidate frames corresponding to the multiple moving objects may be the same frame or different frames; the landing key frames corresponding to the multiple moving objects may be the same frame or different frames.

[0087] In some embodiments, before step S143, it further includes: obtaining first key point information based on the landing key frame. Optionally, obtaining first key point information based on the landing key frame includes: inputting the landing key frame into a preset model to obtain the first key point information.

[0088] In some embodiments, the first key point information may include: heel information, toe information, hip information, head information, and / or arm information. Optionally, the first key point information may further include long jump mat scale information. Here, the heel information may be the coordinates where the heel is located; the toe information may be the coordinates where the toe is located; the hip information may be the distance of the hip from the ground and / or the coordinates mapped on the ground; the head information may be the distance of the head from the ground and / or the coordinates mapped on the ground; the arm information may be information such as whether the arm swings.

[0089] In some embodiments, the preset model may be any model for key point detection. The key points may include human key points and / or long jump mat scale lines, etc.; human key points include the heels of the left and right feet, the toes of the left and right feet, the hips, and / or the arms, etc. For example, the preset model may be a key point detection model; the key point detection model may be a Vision Transformer model, or an HrNet model, or a Vision Transformer + HrNet model, etc.

[0090] In some embodiments, step S143 may include: determining the scale in the long jump mat scale line information corresponding to the heel information after the jump as the long jump distance.

[0091] Moreover, the embodiments of the present invention can measure the long jump distances of multiple moving objects by obtaining the landing key frames of the moving objects and determining the long jump distances of the moving objects based on the key point information of the landing key frames. On the one hand, the implementation is relatively simple, and on the other hand, the long jump distances of multiple moving objects can be obtained separately, which can reduce the interference of the long jump distances of each moving object, etc.

[0092] In some embodiments, the method further includes: when the intersection over union is less than a first threshold and / or the confidence of the human body box is less than a second threshold, outputting a first prompt message; wherein, the first prompt message is used to prompt the moving object to move to the movement area or leave the movement area.

[0093] In some embodiments, outputting the first prompt message includes: outputting the first prompt message by voice and / or displaying the first prompt message. Optionally, the first prompt message can be a warning sound. Optionally, the first prompt message can be voice information or text information such as "Please ask the interfering person to leave" or "Please ask the moving object to move into the range collected by the camera".

[0094] In the embodiments of the present invention, when the intersection over union between the human body box and the long jump mat target box is less than the first threshold and the confidence of the human body box is less than the second threshold, it is highly likely that the person where the human body box is located is not the moving object but the interfering person; thus, by outputting the first prompt message, the interfering person can be warned, which is convenient for accurately obtaining the movement area of the moving object preparing for long jump, etc.

[0095] In some embodiments, step S142 includes: inputting the second video frame into a second classification model to obtain the takeoff key frame and the landing key frame of at least one moving object.

[0096] Optionally, inputting the second video into the second classification model to obtain the takeoff key frame of at least one moving object includes: inputting the second video frame into the second classification model to determine at least one takeoff candidate frame of the moving object; based on at least one landing candidate frame, determining the landing key frame of the moving object. Here, one moving object corresponds to at least one takeoff key frame; in step S142, at least one takeoff candidate frame corresponding to one or more moving objects can be determined, and the takeoff key frame corresponding to one or more moving objects can be determined. Here, when there are multiple moving objects currently, the takeoff candidate frames corresponding to the multiple moving objects can be the same frame or different frames; the takeoff key frames corresponding to the multiple moving objects can be the same frame or different frames.

[0097] As Figure 4 shown, in some embodiments, in step S142, it includes:

[0098] Step S1421: Input the second video frame into the second classification model to obtain the first probability of the takeoff category and the second probability of the landing category of at least one video frame;

[0099] Step S1422: If the first probability of the video frame is greater than or equal to the second probability, determine the video frame as a takeoff candidate frame; or, if the first probability of the video frame is less than the second probability, determine the video frame as a landing candidate frame;

[0100] Step S1423: Determine the takeoff key frame based on at least one takeoff candidate frame;

[0101] Step S1424: Determine the landing key frame based on at least one landing candidate frame.

[0102] Optionally, the first probability is the takeoff class probability; the second probability is the landing class probability.

[0103] Optionally, both the first probability and the second probability can be values greater than or equal to 0 and less than or equal to 1.

[0104] Exemplarily, as Figure 5 shown, the takeoff class probability of the takeoff candidate frame is greater than the takeoff class probability of the landing candidate frame, and the landing class probability of the landing candidate frame is greater than the landing class probability of the takeoff candidate frame; the terminal can determine the video frame corresponding to the maximum value of the takeoff class probability in the takeoff candidate frames as the takeoff key frame, and determine the video frame corresponding to the maximum value of the landing class probability in the landing candidate frames as the landing key frame.

[0105] In the embodiments of the present invention, the takeoff candidate frames and landing candidate frames of the standing long jump can be determined through the probability values belonging to the takeoff class probability and the landing class probability, and the takeoff key frame can be determined from the takeoff candidate frames and the landing key frame can be determined from the landing candidate frames, so as to realize the extraction speed of obtaining the takeoff key frame and the landing candidate frame.

[0106] In some embodiments, S1423 includes one of the following:

[0107] If the difference between the maximum value and the minimum value of the first probability in at least two consecutive takeoff candidate frames is less than or equal to a third threshold, determine the takeoff candidate frame corresponding to the intermediate value of the first probability among the at least two takeoff candidate frames as the takeoff key frame;

[0108] If there is no difference between the first probabilities of two consecutive takeoff candidate frames that is less than or equal to the third threshold, determine the takeoff candidate frame corresponding to the maximum value of the first probability as the takeoff key frame;

[0109] If the first probability of the takeoff candidate frame is greater than or equal to a fourth threshold, determine the takeoff candidate frame as the takeoff key frame.

[0110] Optionally, if the maximum value of the first probability in the takeoff candidate frame is greater than or equal to the fourth threshold, determine the takeoff candidate frame corresponding to the maximum value as the takeoff key frame; or, if the maximum value in the takeoff candidate frame is less than the fourth threshold, determine that there is no takeoff key frame.

[0111] Optionally, there is no difference between the first probabilities of two consecutive takeoff candidate frames that is less than or equal to the third threshold, that is, the difference between the first probabilities of any two consecutive takeoff candidate frames is greater than the third threshold.

[0112] Optionally, the third threshold may be a value less than or equal to 0.1. For example, the third threshold may be 0.02 or 0.05, etc.

[0113] Optionally, the fourth threshold may be a value greater than or equal to 0.5. For example, the fourth threshold may be 0.5 or 0.6, etc.

[0114] Exemplarily, the third threshold is 0.02. The first probabilities of three consecutive takeoff candidate frames obtained by the terminal are 0.90, 0.91, or 0.92 respectively; then the maximum value of the first probabilities of the three takeoff candidate frames is 0.92 and the minimum value is 0.90, and the difference between the maximum value and the minimum value is 0.2. Then, among the first probabilities of the three takeoff candidate frames, the takeoff candidate frame corresponding to the middle value of 0.91 can be selected as the key takeoff frame.

[0115] Exemplarily, the third threshold is 0.02. The terminal obtains 10 takeoff candidate frames; the difference between the first probabilities of any two consecutive takeoff candidate frames among the 10 takeoff candidate frames is greater than 0.02. Then, the takeoff candidate frame corresponding to the maximum value of the first probabilities of the 10 takeoff candidate frames is selected as the key takeoff frame.

[0116] In this way, in the embodiments of the present invention, the takeoff candidate frame with the middle value can be extracted from the takeoff candidate frames as the key takeoff frame, or the takeoff candidate frame with the maximum or relatively large value of the takeoff class probability in the takeoff candidate frames can be used as the key takeoff frame. In this way, the key takeoff frame of the real takeoff posture can be accurately obtained, so as to facilitate the subsequent accurate ranging of the standing long jump. And, the key takeoff frame can be obtained in multiple ways, which can adapt to more application scenarios.

[0117] In some embodiments, step S1424 includes one of the following:

[0118] If the difference between the maximum value and the minimum value of the second probabilities in at least two consecutive landing candidate frames is less than or equal to the fifth threshold, determine the landing candidate frame corresponding to the middle value of the second probabilities in the at least two landing candidate frames as the candidate key frame;

[0119] If there is no difference between the second probabilities of two consecutive landing candidate frames that is less than or equal to the fifth threshold, determine the landing candidate frame corresponding to the maximum value of the second probabilities as the candidate key frame;

[0120] If the second probability of the landing candidate frame is greater than or equal to the sixth threshold, determine the landing candidate frame as the candidate key frame.

[0121] Optionally, if the maximum value of the first probabilities in the landing candidate frame is greater than or equal to the sixth threshold, determine the takeoff candidate frame corresponding to the maximum value as the key takeoff frame; or, if the maximum value of the takeoff candidate frame is less than the sixth threshold, determine that there is no key landing frame.

[0122] Optionally, the difference between the second probabilities of two consecutive landing candidate frames is less than or equal to a fifth threshold, that is, the difference between the second probabilities of any two consecutive landing candidate frames is greater than the fifth threshold.

[0123] Optionally, the fifth threshold may be a value less than or equal to 0.1. For example, the fifth threshold may be 0.02 or 0.05, etc. Optionally, the fifth threshold is equal to the third threshold; or, the fifth threshold is not equal to the third threshold.

[0124] Optionally, the sixth threshold may be a value greater than or equal to 0.5. For example, the sixth threshold may be 0.5 or 0.6, etc. Optionally, the sixth threshold is equal to the fourth threshold; or, the sixth threshold is not equal to the fourth threshold.

[0125] Exemplarily, the fifth threshold is 0.02. The second probabilities of three consecutive landing candidate frames obtained by the terminal are 0.90, 0.91, or 0.92 respectively; then the maximum value of the second probabilities of the three landing candidate frames is 0.92 and the minimum value is 0.90, and the difference between the maximum value and the minimum value is 0.2. Then, among the second probabilities of the three landing candidate frames, the landing candidate frame corresponding to the middle value of 0.91 can be selected as the landing key frame.

[0126] Exemplarily, the fifth threshold is 0.02. The terminal obtains 10 landing candidate frames; the difference between the second probabilities of any two consecutive landing candidate frames among the 10 landing candidate frames is greater than 0.02. Then, the landing candidate frame corresponding to the maximum value of the second probabilities of the 10 landing candidate frames is selected as the landing key frame.

[0127] In this way, in the embodiment of the present invention, the landing key frame can be obtained by extracting the landing candidate frame with the middle value among the landing candidate frames, or by using the landing candidate frame with the maximum value or relatively large value of the landing class probability among the landing candidate frames as the landing key frame. In this way, the landing key frame of the true landing posture can be accurately obtained, so as to facilitate the subsequent accurate ranging of the standing long jump. And, the landing key frame can be obtained in multiple ways, which can adapt to more application scenarios.

[0128] In some embodiments, the method further includes: outputting a second prompt message when the toe tip information in the second key point information of the takeoff key frame exceeds the takeoff line; and / or, outputting a second prompt message when the hip information in the first key point information meets a predetermined condition; wherein, the second prompt message is used to indicate that the long jump is a violation action.

[0129] In some embodiments, the second key point information may include: heel information, toe tip information, hip information, head information, and / or arm information. Optionally, the second key point information may further include long jump mat scale information.

[0130] In some embodiments, outputting the second prompt message includes: outputting the second prompt message by voice and / or displaying the second prompt message. Optionally, the second prompt message may be a warning sound; the warning sound in the second prompt message is different from the warning sound in the first prompt message. Optionally, the second prompt message may be voice information or text information such as "illegal operation" or "the result of this standing long jump is not selected".

[0131] In some embodiments, the buttocks meeting the predetermined condition may be the buttocks touching the ground or the distance between the buttocks and the ground being less than the first distance.

[0132] In the embodiments of the present invention, the terminal can determine that the moving object belongs to an illegal operation by the toe information in the second key point in the takeoff key frame exceeding the takeoff line or the buttocks information in the landing key touching the ground, etc., and thus warn through the second prompt message; in this way, it is possible to facilitate excluding the interference of some illegal actions on the long jump distance and obtain an accurate long jump distance.

[0133] In some examples, the method further includes: obtaining the second key point information based on the takeoff key frame. Optionally, obtaining the second key point information based on the takeoff key frame includes: inputting the takeoff key frame into a preset model to obtain the second key point information.

[0134] In some embodiments, step 16 includes: when the toe information in the second key point information does not exceed the takeoff line and the buttocks information in the first key information does not meet the predetermined condition, determining the long jump distance of the moving object according to the heel information in the first key information and the long jump mat scale information.

[0135] In the embodiments of the present invention, if the toe information in the second key point information does not exceed the takeoff line and the buttocks information in the first key information does not meet the predetermined condition, it is determined that the moving object has not performed an illegal long jump. In this way, at this time, it is determined that the long jump distance of the moving object is a true non-illegal long jump distance, improving the reliability of the long jump result.

[0136] As Figure 6 shown, in some embodiments, step S143 includes:

[0137] Step S1431: Based on the long jump mat scale line information, determine the first scale coordinate and the second scale coordinate closest to the heel information; the distance of the first scale coordinate from the takeoff line is closer than the distance of the second scale coordinate from the takeoff line;

[0138] Step S1432: Determine the first actual coordinate and the second actual coordinate corresponding to the first scale coordinate and the second scale coordinate on the long jump mat respectively;

[0139] Step S1433: Determine the long jump distance based on the first scale coordinate, the second scale coordinate, the first actual coordinate, and the second actual coordinate.

[0140] Optionally, step S1433 includes: determining a first value based on the difference between the second scale coordinate and the first scale coordinate; determining a second value based on the difference between the second actual coordinate and the first actual coordinate; determining a third value based on the ratio of the second value to the first value; determining a fourth value based on the difference between the heel information in the first key information and the first scale coordinate; determining a fifth value based on the product of the fourth value and the third value; and determining the long jump distance based on the sum of the first actual coordinate and the fifth value.

[0141] Exemplarily, as Figure 7 shown, if the first actual coordinate and the second actual coordinate on the long jump mat are 150 centimeters (cm) and 160 cm respectively; the heel information in the landing key frame is (x, y), and the first scale coordinate and the second scale coordinate on the long jump mat scale line closest to the heel information in the landing key frame are (a, y) and (b, y) respectively; then the long jump distance

[0142] Optionally, step S1433 includes: determining a first value based on the difference between the second scale coordinate and the first scale coordinate; determining a second value based on the difference between the second actual coordinate and the first actual coordinate; determining a third value based on the ratio of the second value to the first value; determining a sixth value based on the difference between the second scale coordinate and the heel information in the first key point information; determining a seventh value based on the product of the sixth value and the third value; and determining the long jump distance based on the difference between the second actual coordinate and the seventh value.

[0143] Exemplarily, as Figure 7 shown, if the first actual coordinate and the second actual coordinate on the long jump mat are 150 centimeters (cm) and 160 cm respectively; the heel information in the landing key frame is (x, y), and the first scale coordinate and the second scale coordinate on the long jump mat scale line closest to the heel information in the landing key frame are (a, y) and (b, y) respectively; then the long jump distance

[0144] In this way, in the embodiment of the present invention, the long jump distance can be obtained by detecting the key information in the landing key frame; on the one hand, the purpose of multiple people measuring the distance simultaneously with little interference can be achieved, and on the other hand, there is no need to wear any electronic devices, the hardware cost is simple, and the calculation process is relatively simple.

[0145] In some other embodiments, step S143 includes: determining alternative jump distances corresponding to each landing candidate frame based on at least two landing candidate frames; and determining the jump distance based on the average value of the alternative jump distances corresponding to at least two landing candidate frames.

[0146] Optionally, the method for determining the alternative jump distance corresponding to each landing candidate frame is similar to the method for determining the jump distance based on the key landing frame in step 16, which will not be elaborated here. For example, determining the alternative jump distance corresponding to each landing candidate frame based on at least two landing candidate frames includes: obtaining at least two landing candidate frames; determining a first scale coordinate and a second scale coordinate that are closest to the heel information based on the long jump mat scale line information of each landing candidate frame; the distance of the first scale coordinate from the takeoff line is closer to the distance of the second scale coordinate from the takeoff line; determining the first actual coordinate and the second actual coordinate corresponding to the first scale coordinate and the second scale coordinate on the long jump mat respectively; and determining the alternative jump distance of the landing candidate frame based on the first scale coordinate, the second scale coordinate, the first actual coordinate, and the second actual coordinate.

[0147] In the embodiments of the present invention, the alternative jump distance corresponding to the landing candidate frame can be repeatedly calculated, and the average value of the multiple alternative jump distances can be taken as the jump distance, so as to balance the error of key point detection and finally obtain a more accurate long jump distance.

[0148] To further explain any embodiment of the present invention, a specific embodiment is provided below.

[0149] As Figure 8 shown, the embodiments of the present invention provide a long jump distance measurement method, which is executed by a terminal and includes the following steps:

[0150] Step S21, target detection.

[0151] Optionally, the terminal turns on the camera without setting a fixed long jump position. The moving object can stand still within any range of the camera, facing the side of the camera, with feet together and hands vertically attached to both sides of the body, preparing to perform a long jump. The target detection selects the YOLOX model to identify all people and the long jump mat in the picture, and outputs the long jump mat target box, the human body box, and the confidence level of the human body box. If the confidence level of the human body box is less than 0.5, or the human body box is not on the long jump mat, that is, the intersection over union of the human body box and the long jump mat target box is less than 0.8, the human body box is deleted and a warning for interfering personnel is given; for example, prompting the person to move to a suitable exercise area or leave the exercise area. If the intersection over union of the human body box and the long jump mat target box is greater than or equal to 0.8 and the confidence level of the human body box is greater than or equal to 0.5, it is regarded as a motion area; for example, the area numbers of the motion areas are 1, 2, 3,..., N (N is the total number of motion areas); the N motion areas form a motion area list for subsequent key point detection.

[0152] Here, the moving object can be a tester; the long jump mat can be a standing long jump mat; the camera can be the camera module in the previous embodiments.

[0153] Step S22, key point detection.

[0154] Optionally, after the terminal obtains the result of the motion area list, the moving object starts to move. Key point detection is performed on the moving object and the long jump mat, and the position changes of the key point sequences in each motion area list are recorded. The network backbone of key point detection selects the advantages of both Vision Transformer and HrNet, has the ability of multi-resolution parallelism, and comprehensively captures local and global information, performs excellently in high-resolution dense prediction tasks, and reduces the computational cost. Through the key point detection model, the coordinates of the toes, heels, hips of the moving object, and all scale lines of the long jump mat in each motion area list are recorded for subsequent analysis. A filter is used to filter out the abnormal fluctuations of the key points caused by the unstable output of the key point detection model or the dropped frames of the camera.

[0155] Since both the YOLOX and HrNet models are very large and the time to process one picture is very long, about 700 milliseconds (ms), which cannot meet the real-time requirements in the actual scenario. The YOLOX and HrNet models run and occupy about 6G of video memory, and a single GPU card is usually 32G of video memory, resulting in a waste of GPU resources. Therefore, a distributed architecture of 3 controllers + 3 workers is designed in combination with Kubernetes. One 32G GPU card is allocated to each worker node. In order to maximize the utilization of GPU resources, 5 replicas are published to each worker node, and one object detection service and one key point detection service are started in each replica. Through this architecture, with 3 pieces of 32G GPU resources, 15 services can be started simultaneously, ensuring the high availability of the functions; the FPS (number of pictures processed per second) of key point detection can reach 22, basically meeting the real-time requirements when the video frame rate is 25, and maximizing the utilization of GPU resources.

[0156] Here, the key point information may include the first key point information and / or the second key point information in the previous embodiments. Optionally, step S22 can be after step S23, or step S22 and step S23 can be executed simultaneously.

[0157] Step S23, key pose classification.

[0158] Optionally, according to the movement characteristics of the standing long jump ranging, the terminal pole defines two key postures: the take-off posture and the landing posture. The take-off posture is that the moving object stands in front of the take-off line, swings the arms backward, and prepares to jump; the landing posture is that after the moving object's feet touch the ground, the hip position reaches the lowest point.

[0159] The terminal uses the VGG classification model to construct a deep convolutional neural network by using a series of small-size convolutional kernels of size 3x3 and pooling layers, trains the take-off, and classifies the video frames; the video frame can be the second video frame in the previous embodiment. For example, for each frame of the second video frame, it is input into the network model to obtain the probability values belonging to the take-off category and the landing category. If the take-off category probability of the frame is greater than the landing category probability, then the frame is used as a take-off candidate frame, otherwise, it is used as a landing key candidate frame. The schematic diagram is as Figure 5 shown.

[0160] As Figure 5 shown, in the candidate frame sequence, there are several consecutive frames with very similar probabilities near the frame close to the take-off posture and the landing key posture. Therefore, a method for obtaining the most accurate take-off and landing key postures in the candidate postures with the maximum similarity probability is proposed; for example, a threshold is set to 0.02. If the probability difference between the i-th frame and the j-th frame in the candidate frames is less than the threshold, they are regarded as frames with similar probabilities. The number of frames with specific similar probabilities is counted, and the middle frame is taken as the key posture; if there are no frames with a probability difference less than the threshold, then the frame with the maximum probability is taken as the key posture; where i and j are positive integers respectively. If the probability value of the frame with the highest take-off category probability is greater than 0.5, it is regarded as the take-off posture, otherwise, it is prompted that the take-off posture is not detected; if the probability value of the frame with the highest landing category probability is greater than 0.5, it is regarded as the landing posture, otherwise, it is prompted that the landing posture is not detected. Here, it is specifically pointed out that the key frames of the standing long jump ranging can be quickly located through the probability values of each frame picture belonging to the two categories, realizing a relatively fast extraction speed.

[0161] Here, the take-off category probability and the landing category probability can be the first probability and the second probability in the previous embodiment respectively.

[0162] Step S24, ranging.

[0163] Optionally, the terminal determines whether the moving object has left the camera range or has completed the landing posture. If so, the extraction of key frames is ended. Since the key posture classification model is used, the two key postures of the takeoff posture and the landing posture and the candidate key posture with the highest similarity probability are obtained. When performing human key point detection on the key posture frames, only three points, namely the toes, the heels, and the hips, are used to judge takeoff and landing violations, which greatly reduces the amount of key point annotation and the amount of calculation. It is judged whether the x value of the toes of the takeoff posture exceeds the takeoff line. If it exceeds, a warning of takeoff line violation is given, and the standing long jump score for this time is not counted. It is judged whether the y value of the hips of the landing posture touches the ground. If it touches the ground, a warning of hip touching the ground violation is given, and the standing long jump score for this time is not counted. If there is no violation, the distance is calculated. For the video frames of the landing posture (such as the landing candidate frames or the landing key frames), according to the coordinates of the heel key point (x, y), find the two closest standing long jump mat scale lines (a, y) and (b, y), calculate the ratio of the difference between a and b to the difference between the actual scale line annotation distances, and obtain the scaling ratio of the picture to the actual distance; the schematic diagram is as follows Figure 7 shown; Multiply the distance from x to a by the scaling ratio and add the actual distance of a to obtain the distance of this standing long jump. The calculation formula is as Since the standing long jump is a fast movement, the duration from takeoff to landing generally does not exceed three seconds. The pictures captured by the camera are blurred, resulting in jitter in the recognition of key points. If the distance is judged only for one frame of the landing key frame, the error is relatively large. Therefore, a more accurate ranging method is proposed to range the candidate landing key posture sequence with the highest similarity probability. Select all frames with a probability difference less than the 0.02 threshold from the landing posture frames, repeat the above calculation process, and take the average to balance the error of key point detection. Finally, the distance of the standing long jump is the average value of the calculated distances.

[0164] In the embodiment of the present invention, it is possible to achieve simultaneous ranging of multiple people without pre-dividing a fixed motion area, and it has the characteristics of strong anti-interference.

[0165] Moreover, regarding the detection points for detection, a distributed architecture of 3 controllers (masters) + 3 workers is designed in combination with Kubernetes, which meets the real-time requirements in the actual scenario, makes the best use of resources, and ensures the high availability of functions.

[0166] Moreover, for key pose classification, the concept of the frame with the maximum similarity probability is proposed, which can quickly and accurately extract the key frames for ranging by using the probability values of the classification model. After obtaining the takeoff key frame and / or the landing key frame, it is relatively simple to determine takeoff and / or landing violations by using only three points: the toes, the heels, and the hips. And the distance can be calculated using the ranging model based on the heels and the scale line of the standing long jump mat; moreover, by averaging the distances of the landing key frame and its frame with the maximum similarity probability to balance the error of key point detection, highly accurate ranging is achieved.

[0167] It should be noted here that the following description of the standing long jump ranging device is similar to the description of the above-mentioned standing long jump ranging method, and the beneficial effects of the method will not be elaborated. For the technical details not disclosed in the embodiments of the standing long jump ranging device of the present invention, please refer to the description of the embodiments of the standing long jump ranging method of the present invention.

[0168] As Figure 9 shown, an embodiment of the present invention provides a standing long jump ranging device, including:

[0169] An acquisition module 31, configured to acquire a first video frame;

[0170] A processing module 32, configured to input the first video frame into a first classification model to obtain the confidence of the human body frame of at least one moving object and the intersection over union (IoU) between the human body frame and the target frame of the long jump mat; wherein, the IoU is the ratio of the intersection area to the union area, the intersection area is the overlapping area between the human body frame and the target frame of the long jump mat, and the union area is the union area between the human body frame and the target frame of the long jump mat;

[0171] A determination module 33, configured to determine a list of motion areas of the moving object when the IoU is greater than or equal to a first threshold and the confidence of the human body frame is greater than or equal to a second threshold; wherein, a list of motion areas includes the motion areas of at least one moving object;

[0172] The processing module 32, configured to determine the long jump distance of the moving object based on the list of motion areas.

[0173] In some embodiments, the processing module 32 is configured to acquire a second video frame of at least one moving object jumping in the motion area from the list of motion areas; input the second video frame into a second classification model to obtain the landing key frames of at least one moving object; and determine the long jump distance of the moving object based on the heel information and the scale line information of the long jump mat in the first key point information of the landing key frames.

[0174] In some embodiments, the device further includes: an output module, configured to output a first prompt message when the intersection over union is less than a first threshold and / or the confidence of the human body box is less than a second threshold; wherein, the first prompt message is used to prompt the moving object to move to or leave the movement area.

[0175] In some embodiments, a processing module 32 is configured to input a second video frame into a second classification model to obtain a takeoff key frame and a landing key frame of at least one moving object;

[0176] The output module is configured to output a second prompt message when the toe tip information in the second key point information of the takeoff key frame exceeds the takeoff line; and / or, output a second prompt message when the hip information in the first key point information meets a predetermined condition; wherein, the second prompt message is used to indicate that the long jump is a violation action.

[0177] In some embodiments, the processing module 32 is configured to, when the toe tip information in the second key point information does not exceed the takeoff line and the hip information in the first key information does not meet the predetermined condition, determine the long jump distance of the moving object according to the heel information in the first key information and the long jump mat scale information.

[0178] In some embodiments, the processing module 32 is configured to input a second video frame into a second classification model to obtain a first probability of the takeoff category and a second probability of the landing category of at least one video frame;

[0179] The processing module 32 is further configured to, if the first probability of the video frame is greater than or equal to the second probability, determine the video frame as a takeoff candidate frame; or, if the first probability of the video frame is less than the second probability, determine the video frame as a landing candidate frame; based on at least one takeoff candidate frame, determine the takeoff key frame; based on at least one landing candidate frame, determine the landing key frame.

[0180] In some embodiments, the processing module 32 is configured to perform one of the following:

[0181] If the difference between the maximum value and the minimum value of the first probability in at least two consecutive takeoff candidate frames is less than or equal to a third threshold, determine the takeoff candidate frame corresponding to the intermediate value of the first probabilities of the at least two takeoff candidate frames as the takeoff key frame;

[0182] If there is no difference between the first probabilities of two consecutive takeoff candidate frames that is less than or equal to the third threshold, determine the takeoff candidate frame corresponding to the maximum value of the first probability as the takeoff key frame;

[0183] If the first probability of the takeoff candidate frame is greater than or equal to a fourth threshold, determine the takeoff candidate frame as the takeoff key frame.

[0184] In some embodiments, the processing module 32 is configured to perform one of the following:

[0185] If the difference between the maximum value and the minimum value of the second probability in at least two consecutive landing candidate frames is less than or equal to a fifth threshold, determine the landing candidate frame corresponding to the intermediate value of the second probability among the at least two landing candidate frames as the candidate key frame;

[0186] If there is no difference between the second probabilities of two consecutive landing candidate frames that is less than or equal to the fifth threshold, determine the landing candidate frame corresponding to the maximum value of the second probability as the candidate key frame;

[0187] If the second probability of the landing candidate frame is greater than or equal to a sixth threshold, determine the landing candidate frame as the candidate key frame.

[0188] In some embodiments, the processing module 32 is configured to determine a first scale coordinate and a second scale coordinate that are closest to the heel information based on the long jump mat scale line information; the distance of the first scale coordinate from the takeoff line is closer to the takeoff line than the distance of the second scale coordinate from the takeoff line; determine the first actual coordinate and the second actual coordinate corresponding to the first scale coordinate and the second scale coordinate on the long jump mat respectively;

[0189] The processing module 32 is further configured to determine the long jump distance based on the first scale coordinate, the second scale coordinate, the first actual coordinate, and the second actual coordinate.

[0190] As Figure 10 shown, an embodiment of the present invention further provides a terminal, where the terminal includes a processor 41 and a memory 42 for storing a computer program that can run on the processor 41; wherein, when the processor 41 is used to run the computer program, the standing long jump distance measurement method according to any embodiment of the present invention is implemented.

[0191] In some embodiments, the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchlink dynamic random access memory (SLDRAM), and direct rambus random access memory (DRRAM). The memory of the systems and methods described herein is intended to include, but not be limited to, these and any other suitable types of memory.

[0192] The processor may be an integrated circuit chip with the ability to process signals. In the implementation process, each step of the above method can be completed by the integrated logic circuit in the hardware of the processor or instructions in the form of software. The above-mentioned processor may be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present invention can be directly embodied as being executed and completed by a hardware decoding processor, or executed and completed by a combination of hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method.

[0193] In some embodiments, the embodiments described herein can be implemented using hardware, software, firmware, middleware, microcode, or a combination thereof. For a hardware implementation, the processing unit can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions described in the invention, or a combination thereof.

[0194] For a software implementation, the techniques described herein can be implemented by modules (e.g., procedures, functions, etc.) that execute the functions described herein. The software code can be stored in a memory and executed by a processor. The memory can be implemented inside or outside the processor.

[0195] An embodiment of the present invention provides a computer storage medium. The computer-readable storage medium stores an executable program, and when the executable program is executed by a processor, the steps of the standing long jump distance measurement method according to any embodiment of the present invention can be implemented.

[0196] In some embodiments, the computer storage medium may include: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc.

[0197] It should be noted that: the technical solutions described in the embodiments of the present invention can be arbitrarily combined without conflict.

[0198] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A standing long jump ranging method, characterized in that, The method includes: Obtain a first video frame; Input the first video frame into a first classification model to obtain the confidence of the human body box of at least one moving object and the intersection over union (IoU) between the human body box and the target box of the long jump mat; wherein, the IoU is the ratio of the intersection area to the union area, the intersection area is the overlapping area between the human body box and the target box of the long jump mat, and the union area is the union area between the human body box and the target box of the long jump mat; When the IoU is greater than or equal to a first threshold and the confidence of the human body box is greater than or equal to a second threshold, determine a list of movement areas of the moving object; wherein, one list of movement areas includes the movement areas of at least one of the moving objects; Based on the list of movement areas, determine the long jump distance of the moving object.

2. The method according to claim 1, characterized in that, The determining the long jump distance of the moving object based on the list of movement areas includes: Obtain a second video frame of at least one of the moving objects jumping in the movement area from the list of movement areas; Input the second video frame into a second classification model to obtain the landing key frames of at least one of the moving objects; Based on the heel information in the first key point information of the landing key frame and the long jump mat scale line information, determine the long jump distance of the moving object.

3. The method according to claim 1 or 2, characterized in that, The method further includes: When the IoU is less than the first threshold and / or the confidence of the human body box is less than the second threshold, output a first prompt message; wherein, the first prompt message is used to prompt the moving object to move to or leave the movement area.

4. The method according to claim 2, wherein The inputting the second video frame into a second classification model to obtain the landing key frames of at least one of the moving objects includes: inputting the second video frame into a second classification model to obtain the takeoff key frames and landing key frames of at least one of the moving objects; The method further includes: when the toe tip information in the second key point information of the takeoff key frame exceeds the takeoff line, output a second prompt message; and / or, when the hip information in the first key point information meets a predetermined condition, output the second prompt message; wherein, the second prompt message is used to indicate that the long jump is a foul action.

5. The method according to claim 4, wherein The determining the long jump distance of the moving object based on the heel information in the first key point information of the landing key frame and the long jump mat scale line information includes: When the toe tip information in the second key point information does not exceed the takeoff line and the hip information in the first key information does not meet the predetermined condition, determine the long jump distance of the moving object according to the heel information in the first key information and the long jump mat scale information.

6. The method according to claim 4, wherein The inputting the second video frame into a second classification model to obtain the takeoff key frames and landing key frames of at least one of the moving objects includes: Input the second video frame into a second classification model to obtain a first probability of the takeoff category and a second probability of the landing category of at least one video frame; If the first probability of the video frame is greater than or equal to the second probability, determine the video frame as a takeoff candidate frame; or, if the first probability of the video frame is less than the second probability, determine the video frame as a landing candidate frame; Based on at least one of the takeoff candidate frames, determine the takeoff key frame; Based on at least one of the landing candidate frames, determine the landing key frame.

7. The method according to claim 6, characterized in that, The determining the takeoff key frame based on at least one of the takeoff candidate frames includes one of the following: If the difference between the maximum value and the minimum value of the first probability in at least two consecutive takeoff candidate frames is less than or equal to a third threshold, determine the takeoff candidate frame corresponding to the intermediate value of the first probabilities of the at least two takeoff candidate frames as the takeoff key frame; If there is no difference between the first probabilities of two consecutive takeoff candidate frames that is less than or equal to the third threshold, determine the takeoff candidate frame corresponding to the maximum value of the first probability as the takeoff key frame; If the first probability of the takeoff candidate frame is greater than or equal to a fourth threshold, determine the takeoff candidate frame as the takeoff key frame.

8. The method according to claim 6, wherein The determining the landing key frame based on at least one of the landing candidate frames includes one of the following: If the difference between the maximum value and the minimum value of the second probability in at least two consecutive landing candidate frames is less than or equal to a fifth threshold, determine the landing candidate frame corresponding to the intermediate value of the second probabilities of the at least two landing candidate frames as the candidate key frame; If there is no difference between the second probabilities of two consecutive landing candidate frames that is less than or equal to the fifth threshold, determine the landing candidate frame corresponding to the maximum value of the second probability as the candidate key frame; If the second probability of the landing candidate frame is greater than or equal to a sixth threshold, determine the landing candidate frame as the candidate key frame.

9. The method according to claim 2, characterized in that, The determining the long jump distance of the moving object based on the heel information and the long jump mat scale line information in the first key point information of the landing key frame includes: Based on the long jump mat scale line information, determine the first scale coordinate and the second scale coordinate that are closest to the heel information; the distance of the first scale coordinate from the takeoff line is closer to the takeoff line than the distance of the second scale coordinate from the takeoff line; Determine the first actual coordinate and the second actual coordinate corresponding to the first scale coordinate and the second scale coordinate on the long jump mat respectively; Based on the first scale coordinate, the second scale coordinate, the first actual coordinate and the second actual coordinate, determine the long jump distance.

10. A standing long jump distance measuring device, characterized in that, Includes: An acquisition module for acquiring a first video frame; A processing module for inputting the first video frame into a first classification model to obtain the confidence of the human body frame of at least one moving object and the intersection over union between the human body frame and the long jump mat target frame; wherein, the intersection over union is the ratio of the intersection area to the union area, the intersection area is the overlapping area between the human body frame and the long jump mat target frame, and the union area is the union area between the human body frame and the long jump mat target frame; A determination module, configured to determine a list of movement areas of the moving object when the intersection over union is greater than or equal to a first threshold and the confidence of the human body box is greater than or equal to a second threshold; wherein, one list of movement areas includes the movement areas of at least one of the moving objects. The processing module is configured to determine the long jump distance of the moving object based on the list of movement areas.

11. A terminal, characterized in that, The terminal includes a processor and a memory for storing a computer program that can run on the processor; wherein, when the processor is used to run the computer program, the standing long jump distance measurement method according to any one of claims 1 to 9 is implemented.

12. A computer storage medium, characterized in that, There are computer-executable instructions in the computer storage medium, characterized in that the computer-executable instructions are executed by a processor to implement the standing long jump distance measurement method according to any one of claims 1 to 9.

13. A computer program product, the computer program product comprising a computer program or instructions, characterized in that, When the computer program or instruction is executed by a processor, the standing long jump distance measurement method according to any one of claims 1 to 9 is implemented.

Images