Method and device for judging volleyball passing and padding training based on visual technology

Abstract

The invention discloses a method and device for judging volleyball passing and padding training based on a visual technology, and relates to the technical field of intelligent sports. The method comprises the following steps: a deep learning model detection process: performing real-time frame detection on a target frame and a finger key point; in the volleyball passing and padding judgment process, whether passing and padding are successful or not is judged according to the relation between the key point coordinates and the target frame coordinates; in the process of logical judgment of the mat ball, accurate counting of successful times of mat ball conveying can be realized by setting a motion state mark and mat ball conveying judgment logic. According to the method and the device for judging volleyball dribbling training based on the visual technology, real-time accurate counting is realized by combining dribbling state control based on the computer visual technology, and the fairness and the accuracy of an examination result are improved.

Description

Technical Field

[0001] The present invention relates to the field of smart sports technology, and in particular to a method and device for judging volleyball pass and cushion training based on visual technology. Background Art

[0002] Passing is one of the key techniques in volleyball. Being able to accurately judge the results of passing can help athletes and coaches better understand and improve techniques, and enhance athletes' performance and competitiveness. Passing requires fast movement, flexible turns, and accurate passing of the ball, which can improve teenagers' physical coordination, sensitivity, and reaction speed. Passing requires teenagers to cooperate with their teammates, which can cultivate their teamwork spirit and communication skills. Through passing, teenagers can improve their self-confidence and self-esteem, and enhance their sense of competition and spirit of challenge.

[0003] Since the judgment and counting rules of pass and cushion are relatively complicated, current volleyball pass and cushion training generally adopts the manual counting method. However, it requires long-term concentration, the referee's work intensity is high, and he is prone to fatigue and distraction. The referee is prone to misjudgment, omission of counts, and incorrect counting. If used in examinations, it will affect the fairness and accuracy of the test results. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method and device for judging volleyball pass and cushion training based on visual technology, and to achieve real-time and accurate counting based on computer vision technology combined with pass and cushion state control, thereby improving the fairness and accuracy of the test results.

[0005] In a first aspect, the present invention provides a method for judging volleyball pass and cushion training based on visual technology, comprising:

[0006] Deep learning model detection process: The target frame and finger key points are detected in real-time frames using a deep learning model. The target frame includes a person frame, an arm frame, a hand frame, and a ball frame, and the top of the person frame is the vertex of the head.

[0007] Volleyball pass and cushion judgment process:

[0008] Determine whether the hands are above the head based on the relationship between the coordinates of the finger key points and the coordinates of the human frame;

[0009] When both hands are above the head, it is detected that the IOU between the hand frame and the ball frame is greater than the first set value, and it is judged that the hand touches the ball, and then the ball flies upward, which is judged as a successful pass;

[0010] When both hands are not above the head, it is detected that the IOU of the arm frame and the ball frame is greater than the first set value and the IOU of the two hand frames is greater than the second set value, and the ball flies upward above the head, then it is determined that the pass is successful;

[0011] The logical judgment process of passing the ball:

[0012] After the start signal is sent, the image data of each frame is detected in real time and the motion state flag is set to the initial value, and the number of successful passes is set to 0;

[0013] After detecting that the pass is successful, read the value of the motion state flag: if it is the initial value, set the motion state flag to the first state value; if it is the second state value, set the motion state flag to the first state value, increase the number of successful pass and pass by 1, and then set the motion state flag to the initial value; if it is the first state value, continue to detect;

[0014] After detecting that the pass is successful, read the value of the motion state flag: if it is the initial value, set the motion state flag to the second state value; if it is the first state value, set the motion state flag to the second state value, increase the number of successful passes by 1, and then set the motion state flag to the initial value; if it is the second state value, continue to detect;

[0015] If the ball falls to the ground or the timing is up, the motion state flag is set to the initial value and the training ends; when the training ends, the detection is stopped and the count result of the number of successful passes and cushions is output.

[0016] In a second aspect, the present invention provides a device for judging volleyball pass and cushion training based on visual technology, comprising:

[0017] A deep learning model detection module is used to detect a target frame and finger key points in real time through a deep learning model, wherein the target frame includes a person frame, an arm frame, a hand frame and a ball frame, and the top of the person frame is the vertex of the head;

[0018] Volleyball pass and cushion judgment module, used for:

[0019] Determine whether the hands are above the head based on the relationship between the coordinates of the finger key points and the coordinates of the human frame;

[0020] When both hands are above the head, it is detected that the IOU between the hand frame and the ball frame is greater than the first set value, and it is judged that the hand touches the ball, and then the ball flies upward, which is judged as a successful pass;

[0021] When both hands are not above the head, it is detected that the IOU of the arm frame and the ball frame is greater than the first set value and the IOU of the two hand frames is greater than the second set value, and the ball flies upward above the head, then it is determined that the pass is successful;

[0022] The pass and cushion logic judgment module is used to:

[0023] After the start signal is sent, the image data of each frame is detected in real time and the motion state flag is set to the initial value, and the number of successful passes is set to 0;

[0024] After detecting that the pass is successful, read the value of the motion state flag: if it is the initial value, set the motion state flag to the first state value; if it is the second state value, set the motion state flag to the first state value, increase the number of successful pass and pass by 1, and then set the motion state flag to the initial value; if it is the first state value, continue to detect;

[0025] After detecting that the pass is successful, read the value of the motion state flag: if it is the initial value, set the motion state flag to the second state value; if it is the first state value, set the motion state flag to the second state value, increase the number of successful passes by 1, and then set the motion state flag to the initial value; if it is the second state value, continue to detect;

[0026] If the ball falls to the ground or the timing is up, the motion state flag is set to the initial value and the training ends; when the training ends, the detection is stopped and the count result of the number of successful passes and cushions is output.

[0027] The technical solution provided in the embodiments of the present invention has at least the following technical effects:

[0028] Based on computer vision technology, including target frame IOU algorithm, key point coordinate judgment, etc., combined with pass and cushion state control, real-time and accurate counting is achieved, which improves the fairness and accuracy of the test results and reduces the possibility of human error.

[0029] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The present invention will be further described below in conjunction with embodiments with reference to the accompanying drawings.

[0031] Figure 1 is a flow chart of the method in Embodiment 1 of the present invention;

[0032] Figure 2 It is a schematic diagram of the structure of the device in the second embodiment of the present invention. DETAILED DESCRIPTION

[0033] The embodiments of the present application provide a method and device for judging volleyball pass and cushion training based on visual technology, and realize real-time and accurate counting based on computer vision technology combined with pass and cushion state control, thereby improving the fairness and accuracy of the test results.

[0034] The technical solution in the embodiment of the present application has the following general idea:

[0035] The present invention is based on computer vision technology, and uses a deep convolutional neural network model to identify the key points of people, volleyballs, arms, and fingers, and then judges the pass, pad, ball height, etc. according to business logic to determine whether it is a legal pass or pad action. If there are two consecutive successful passes and pad actions, the number of successful passes and pads is increased by 1, and so on until the end of the assessment. The overall process includes three processes: a deep learning model detection process, real-time frame detection of target frames and finger key points; a volleyball pass and pad judgment process, which determines whether the pass and pad are successful through the relationship between the key point coordinates and the target frame coordinates; and a pass and pad logic judgment process, which can accurately count the number of successful passes and pads by setting the motion state flag and the pass and pad judgment logic.

[0036] Before introducing the specific embodiment, the counting rules of the pass and cushion ball are introduced as follows:

[0037] After hearing the "start" signal, the candidate starts with a self-throwing ball, and continuously completes the self-passing and self-padding actions over the head in the manner of "pad→pass" or "pass→pad" alternating touching the ball. Each set of passing and padding alternating actions is counted as one, and the score corresponding to the number of completed actions within 60 seconds is included in the total score. During the test, the candidate can make adjustments through continuous passing or padding. When adjusting the ball, the clock does not stop, and the count does not count until the action of "pad→pass" or "pass→pad" alternating touching the ball occurs, and then the count continues. When padding the ball, the forearms of both hands must touch the ball at the same time, and the padding height must be such that the entire ball exceeds the top of the head; when passing the ball, both hands must touch the ball at the same time above the forehead to complete the passing action. If there is an interruption in the process of completing the action, such as the ball falling to the ground, holding the ball with the hand, etc., the action is stopped, which is considered one test.

[0038] Pass definition: Raise both hands above the head to make contact with the ball and hit the ball upward.

[0039] Definition of a pass: The forearms of both hands must touch the ball at the same time, and the height of the pass must be such that the entire ball exceeds the top of your head.

[0040] Each candidate has two attempts to take the test, and the best score will be counted in the total score. If a candidate fails to get a score twice due to violations, he / she can take the test again.

[0041] Venue Description:

[0042] One camera, with a height of 2.5 meters, shoots from the side of the candidate.

[0043] Embodiment 1

[0044] This embodiment provides a method for judging volleyball pass and cushion training based on visual technology, such as Figure 1As shown, the following steps may be included:

[0045] S1. Deep learning model detection process: The target frame and finger key points are detected in real-time frames through the deep learning model, wherein the target frame includes a person frame, an arm frame, a hand frame and a ball frame, and the top of the person frame is the vertex of the head (the person frame includes at least the head, and may also include the head and the entire body).

[0046] The target frame can be represented as the coordinates of the upper left corner (including x-coordinate and y-coordinate) and the length and width information of the frame; the finger key point is represented as the x-coordinate and y-coordinate of the point; and whether the hands are above the head is determined based on the relationship between the y-coordinate of the finger key point and the y-coordinate of the upper left corner of the human frame.

[0047] Detecting finger key points specifically involves detecting the top, bottom, and two joints of each finger, a total of four points. The image dataset will be input into the deep convolutional neural network model for training according to the detection target to improve the recognition accuracy.

[0048] S2. Volleyball pass and cushion judgment process:

[0049] Whether the hands are above the head is determined based on the relationship between the coordinates of the key points of the fingers and the coordinates of the human frame; specifically, based on the relationship between the coordinates of the key points of the fingers and the coordinates of the upper left corner of the human frame, if the y coordinate of the finger is above the y coordinate of the upper left corner of the human frame, it is determined that the hands are above the head.

[0050] When both hands are above the head, it is detected that the IOU of the hand frame and the ball frame is greater than a first set value (for example, 0.2), and it is judged that the hand touches the ball (IOU is the intersection-over-union ratio, which is the intersection of the hand frame area and the ball frame area divided by the union of the hand frame area and the ball frame area), and then the ball flies upward, which is judged as a successful pass;

[0051] When both hands are not above the head, it is detected that the IOU of the arm frame and the ball frame is greater than the first set value (for example, 0.2, used to determine whether the ball is touched) and the IOU of the two hand frames is greater than the second set value (for example, 0.8, used to determine whether the two hands overlap). When the ball flies upward above the head (the y coordinate of the ball is above the y coordinate of the upper left corner of the human frame), it is judged that the pass is successful.

[0052] S3, pass and cushion logic judgment process:

[0053] After the start signal is sent, the image data of each frame is detected in real time and the motion state flag is set to the initial value, and the number of successful passes is set to 0;

[0054] After detecting that the pass is successful, read the value of the motion state flag: if it is the initial value, set the motion state flag to the first state value; if it is the second state value, set the motion state flag to the first state value, increase the number of successful pass and pass by 1, and then set the motion state flag to the initial value; if it is the first state value, continue to detect;

[0055] After detecting that the pass is successful, read the value of the motion state flag: if it is the initial value, set the motion state flag to the second state value; if it is the first state value, set the motion state flag to the second state value, increase the number of successful passes by 1, and then set the motion state flag to the initial value; if it is the second state value, continue to detect;

[0056] If the ball falls to the ground or the timing is up, the motion state flag is set to the initial value and the training ends; when the training ends, the detection is stopped and the count result of the number of successful passes and cushions is output.

[0057] The program logic for counting the number of successful passes is as follows:

[0058] At the beginning, the motion state flag is set to the initial value, and the number of successful passes is set to 0;

[0059] If it is determined that the ball is successfully blocked, the motion state flag is set to a first state value (for example, 1);

[0060] If the pass is determined to be successful, the motion state flag is set to a second state value (e.g., 2);

[0061] When the first state value and the second state value appear successively (1 appears first and then 2, or 2 appears first and then 1), the number of successful passes and cushions is increased by 1, and then the motion state flag is restored to the initial value 0.

[0062] The logical process is clear, the state changes are clear, and there will be no process confusion.

[0063] In a specific embodiment, the examination process is as follows:

[0064] Step 1. After sending the "start" signal, start to detect the image data of each frame in real time, set the motion status flag status to 0 initial value, and set the "pass and cushion" success number flag count to 0 initial value.

[0065] Status value description: 0 means initial, 1 means successful pass (i.e. first status value), 2 means successful pass (i.e. second status value).

[0066] Step 2. After the candidate throws the ball, no matter it is the first "pad" or "pass", as long as one of the actions is detected to be successful, the test is considered to have started. The following will assume that the first time is a successful "pad".

[0067] Step 3. After the first judgment that the "ball-passing" action is successful, the motion status flag status is set to 1.

[0068] Step 4. The process continues. After the "ball pad" action is successful, the ball's trajectory is first upward and then falling. The judgment branch of the ball falling is as follows:

[0069] Judgment branch one: After the ball falls, if a successful "pass" action is detected, and the status flag value is 1, it means that the last time was a successful "blocking" action, and this time is a successful "passing" action. Both actions are successfully completed at the same time, so the success count flag count is increased by 1, and then the motion status flag status is set to 0.

[0070] Judgment branch two: If a successful "pad" action is detected after the ball falls, no processing is done and the process continues.

[0071] Judgment branch three: If the ball lands on the ground after it falls, the status flag value is set to the initial value 0 and the test ends.

[0072] Step 5. If the ball does not stop, the process repeats the above steps until the timer expires and the assessment ends. When the counting results are output, the entire video is saved and the action analysis and suggestions are output.

[0073] Computer vision technology can be used to monitor and analyze athletes' passing and cushioning movements in real time, improve athletes' training efficiency, and provide detailed data analysis and feedback to help athletes better adjust and improve their techniques, and help students understand their own techniques and movements and improve their athletic level.

[0074] Based on the same inventive concept, the present application also provides a device corresponding to the method in Example 1, see Example 2 for details.

[0075] Embodiment 2

[0076] In this embodiment, a device for judging volleyball pass and cushion training based on visual technology is provided. Figure 2 As shown, including:

[0077] A deep learning model detection module is used to detect a target frame and finger key points in real time through a deep learning model, wherein the target frame includes a person frame, an arm frame, a hand frame and a ball frame, and the top of the person frame is the vertex of the head;

[0078] Volleyball pass and cushion judgment module, used for:

[0079] Determine whether the hands are above the head based on the relationship between the coordinates of the finger key points and the coordinates of the human frame;

[0080] When both hands are above the head, it is detected that the IOU between the hand frame and the ball frame is greater than the first set value, and it is judged that the hand touches the ball, and then the ball flies upward, which is judged as a successful pass;

[0081] When both hands are not above the head, it is detected that the IOU of the arm frame and the ball frame is greater than the first set value and the IOU of the two hand frames is greater than the second set value, and the ball is flying upward, then it is determined that the pass is successful;

[0082] The pass and cushion logic judgment module is used to:

[0083] After the start signal is sent, the image data of each frame is detected in real time and the motion state flag is set to the initial value, and the number of successful passes is set to 0;

[0084] After detecting that the pass is successful, read the value of the motion state flag: if it is the initial value, set the motion state flag to the first state value; if it is the second state value, set the motion state flag to the first state value, increase the number of successful pass and pass by 1, and then set the motion state flag to the initial value; if it is the first state value, continue to detect;

[0085] After detecting that the pass is successful, read the value of the motion state flag: if it is the initial value, set the motion state flag to the second state value; if it is the first state value, set the motion state flag to the second state value, increase the number of successful passes by 1, and then set the motion state flag to the initial value; if it is the second state value, continue to detect;

[0086] If the ball falls to the ground or the timing is up, the motion state flag is set to the initial value and the training ends; when the training ends, the detection is stopped and the count result of the number of successful passes and cushions is output.

[0087] Specifically, the target frame is represented by the coordinates of the upper left corner and the length and width information of the frame, and the finger key point is represented by the x-coordinate and y-coordinate of the point. The relationship between the y-coordinate of the finger key point and the y-coordinate of the upper left corner of the human frame is used to determine whether the hands are above the head.

[0088] It also includes that a camera for acquiring images is set on the side of the user; when the counting results are output, the whole video is saved, and the action analysis and suggestions are output at the same time.

[0089] Preferably, detecting the key points of the fingers specifically involves detecting the top end, the bottom end and two joint points of each finger.

[0090] Specifically, the first setting value is 0.2, and the second setting value is 0.8.

[0091] Since the device introduced in the second embodiment of the present invention is a device used to implement the method of the first embodiment of the present invention, those skilled in the art can understand the specific structure and deformation of the device based on the method introduced in the first embodiment of the present invention, so it is not described here in detail. All devices used in the method of the first embodiment of the present invention belong to the scope of protection of the present invention.

[0092] The present invention is based on computer vision technology, including target frame IOU algorithm, key point coordinate judgment, etc., and combines pass and cushion state control to achieve real-time accurate counting, thereby improving the fairness and accuracy of test results, reducing the possibility of human error, reducing the workload of referees, and avoiding referee fatigue and distraction. The technical solution of the present invention has the following improvements: 1. Improving the fairness and accuracy of test results and reducing the possibility of human error; 2. Through computer vision technology, athletes' pass and cushion movements can be monitored and analyzed in real time to improve athletes' training efficiency; 3. Computer vision technology can provide detailed data analysis and feedback, thereby helping athletes to better adjust and improve their techniques, helping students understand their techniques and movements, and improving their athletic level.

[0093] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0094] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0095] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0096] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0097] Although the specific implementation modes of the present invention are described above, those skilled in the art should understand that the specific implementation modes described are only illustrative and are not intended to limit the scope of the present invention. Equivalent modifications and changes made by those skilled in the art in accordance with the spirit of the present invention should be included in the scope of protection of the claims of the present invention.

Claims

1. A method for judging volleyball pass and cushion training based on visual technology, characterized in that: include: Deep learning model detection process: The target frame and finger key points are detected in real-time frames using a deep learning model. The target frame includes a person frame, an arm frame, a hand frame, and a ball frame, and the top of the person frame is the vertex of the head. Volleyball pass and cushion judgment process: Determine whether the hands are above the head based on the relationship between the coordinates of the finger key points and the coordinates of the human frame; When both hands are above the head, it is detected that the IOU between the hand frame and the ball frame is greater than the first set value, and it is judged that the hand touches the ball, and then the ball flies upward, which is judged as a successful pass; When both hands are not above the head, it is detected that the IOU of the arm frame and the ball frame is greater than the first set value and the IOU of the two hand frames is greater than the second set value, and the ball flies upward above the head, then it is determined that the pass is successful; The logical judgment process of passing the ball: After the start signal is sent, the image data of each frame is detected in real time and the motion state flag is set to the initial value, and the number of successful passes is set to 0; After detecting that the pass is successful, read the value of the motion state flag: if it is the initial value, set the motion state flag to the first state value; if it is the second state value, set the motion state flag to the first state value, increase the number of successful pass and pass by 1, and then set the motion state flag to the initial value; if it is the first state value, continue to detect; After detecting that the pass is successful, read the value of the motion state flag: if it is the initial value, set the motion state flag to the second state value; if it is the first state value, set the motion state flag to the second state value, increase the number of successful passes by 1, and then set the motion state flag to the initial value; if it is the second state value, continue to detect; If the ball falls to the ground or the timing is up, the motion state flag is set to the initial value and the training ends; when the training ends, the detection is stopped and the count result of the number of successful passes and cushions is output.

2. The method according to claim 1, characterized in that: The target frame is represented by the coordinates of the upper left corner and the length and width of the frame. The finger key point is represented by the x-coordinate and y-coordinate of the point. The relationship between the y-coordinate of the finger key point and the y-coordinate of the upper left corner of the human frame is used to determine whether the hands are above the head.

3. The method according to claim 1 or 2, characterized in that: Detecting finger key points specifically involves detecting the top, bottom, and two joint points of each finger.

4. The method according to claim 1, characterized in that: The camera used to acquire images is set on the side of the user; when the counting results are output, the entire video is saved, and action analysis and suggestions are output at the same time.

5. The method according to claim 1, characterized in that: The first setting value is 0.2, and the second setting value is 0.

8.

6. A device for judging volleyball pass and cushion training based on visual technology, characterized in that: include: A deep learning model detection module is used to detect a target frame and finger key points in real time through a deep learning model, wherein the target frame includes a person frame, an arm frame, a hand frame and a ball frame, and the top of the person frame is the vertex of the head; Volleyball pass and cushion judgment module, used for: Determine whether the hands are above the head based on the relationship between the coordinates of the finger key points and the coordinates of the human frame; When both hands are above the head, it is detected that the IOU between the hand frame and the ball frame is greater than the first set value, and it is judged that the hand touches the ball, and then the ball flies upward, which is judged as a successful pass; When both hands are not above the head, it is detected that the IOU of the arm frame and the ball frame is greater than the first set value and the IOU of the two hand frames is greater than the second set value, and the ball flies upward above the head, then it is determined that the pass is successful; The pass and cushion logic judgment module is used to: After the start signal is sent, the image data of each frame is detected in real time and the motion state flag is set to the initial value, and the number of successful passes is set to 0; After detecting that the pass is successful, read the value of the motion state flag: if it is the initial value, set the motion state flag to the first state value; if it is the second state value, set the motion state flag to the first state value, increase the number of successful pass and pass by 1, and then set the motion state flag to the initial value; if it is the first state value, continue to detect; After detecting that the pass is successful, read the value of the motion state flag: if it is the initial value, set the motion state flag to the second state value; if it is the first state value, set the motion state flag to the second state value, increase the number of successful passes by 1, and then set the motion state flag to the initial value; if it is the second state value, continue to detect; If the ball falls to the ground or the timing is up, the motion state flag is set to the initial value and the training ends; when the training ends, the detection is stopped and the count result of the number of successful passes and cushions is output.

7. The device according to claim 6, characterized in that: The target frame is represented by the coordinates of the upper left corner and the length and width of the frame. The finger key point is represented by the x-coordinate and y-coordinate of the point. The relationship between the y-coordinate of the finger key point and the y-coordinate of the upper left corner of the human frame is used to determine whether the hands are above the head.

8. The device according to claim 6 or 7, characterized in that: Detecting finger key points specifically involves detecting the top, bottom, and two joint points of each finger.

9. The device according to claim 6, characterized in that: The camera used to acquire images is set on the side of the user; when the counting results are output, the entire video is saved, and action analysis and suggestions are output at the same time.

10. The device according to claim 6, characterized in that: The first setting value is 0.2, and the second setting value is 0.8.

Images