Tennis training assistance system and method based on tennis videos

Abstract

Disclosed in the present invention is a tennis training assistance system based on tennis videos. The system comprises a system cloud platform, which is internally provided with a video collection unit, a video processing and analysis unit, a training feedback unit, a training suggestion and personalized training unit, and a training simulation unit, wherein the video collection unit is internally provided with a multi-angle-of-view high-definition camera and a high-speed camera; the video processing and analysis unit is internally provided with a video processing module, a video analysis module and a computer vision module; and the training suggestion and personalized training unit is internally provided with a machine learning algorithm module, a deep learning algorithm module, a suggestion generation module and a personalized training formulation module. The present invention can capture and analyze key data in real time, such as the actions of athletes, ball path trajectories and ball striking points, and can provide personalized training suggestions. The present invention has the advantages of real-time performance, precision, personalization, etc., and can significantly improve the scientific rigor and effectiveness of tennis training.

Description

A tennis training auxiliary system and method based on tennis video TECHNICAL FIELD

[0001] The present application relates to the technical field of sports training, in particular to a tennis training auxiliary system and method based on tennis video. BACKGROUND

[0002] Tennis is a ball game, usually between two single players or two pairs. Players hit the ball with a racket across the net on the tennis court, the goal is to make the opponent unable to effectively return the ball, the player who cannot return the ball will lose points, and the opponent player scores. Tennis is a very attractive sport with unique sports characteristics and deep connotation. With the popularity of tennis, athletes and coaches have higher and higher requirements for the scientificity and accuracy of training.

[0003] Traditional tennis training mainly relies on the experience and subjective judgment of coaches, and lacks objective data support. In recent years, video analysis technology has gradually increased in sports training, but the existing tennis training auxiliary system mostly only provides simple video playback function, lacks deep data analysis and real-time feedback capability. Therefore, it has important practical significance to design a tennis training auxiliary system based on tennis video. SUMMARY

[0004] The purpose of the present application is to provide a tennis training auxiliary system based on tennis video to solve the problems raised in the background.

[0005] To achieve the above purpose, the present application provides the following technical scheme: a tennis training auxiliary system based on tennis video, comprising a system cloud platform, the inside of the system cloud platform comprising a video acquisition unit, a video processing and analysis unit, a training feedback unit, a training suggestion and individualized training unit, and a simulation training unit, the inside of the video acquisition unit comprising a multi-view high-definition camera and a high-speed camera, the inside of the video processing and analysis unit comprising a video processing module, a video analysis module, and a computer vision module, the inside of the training feedback unit being provided with a real-time feedback module, and the inside of the training suggestion and individualized training unit comprising a machine learning algorithm module, a deep learning algorithm module, a suggestion generation module, and an individualized training formulation module.

[0006] Preferably, the system cloud platform saves system data in a cloud database in real time through a wireless transmission module.

[0007] Preferably, the multi-view high-definition camera and high-speed camera capture the training video of the athlete in real time, and the multi-view high-definition camera video acquisition unit collects data of the multi-view high-definition camera and high-speed camera through a video acquisition module.

[0008] Preferably, the multi-view high-definition camera is installed at different positions around the court to ensure full coverage of the movement of the players and the trajectory of the ball, and the high-speed camera is installed on the electric pan-tilt head on the surface of the annular track in the court.

[0009] Preferably, the video processing module is used for feature extraction, classification identification and image correction of video images.

[0010] Preferably, the video analysis module performs action capture and identification, ball trajectory analysis and hitting point detection through the computer vision module.

[0011] Preferably, the machine learning algorithm module includes decision tree algorithm, random forest algorithm, support vector machine algorithm and neural network algorithm, the deep learning algorithm module includes convolutional neural network and recurrent neural network, and the suggestion generation module and the individualized training formulation module generate training suggestion content and individualized training content based on the machine learning algorithm module and the deep learning algorithm module respectively.

[0012] Preferably, the simulation training unit includes virtual reality equipment and augmented reality equipment.

[0013] The application discloses a tennis training auxiliary method based on tennis video, which comprises the following steps: first, capturing the training video of the player in real time through a multi-angle high-definition camera and a high-speed camera; the multi-angle high-definition camera can be installed at different positions around the court to ensure that the player's movements and the movement trajectory of the ball are covered in all directions; the high-speed camera is installed on an electric pan-tilt head on the surface of a ring-shaped track, so that the high-speed camera can move in a ring shape, and multi-directional and dead-angle-free shooting and recording are facilitated, and subsequent multi-dimensional analysis is facilitated; then, the data of the multi-angle high-definition camera and the high-speed camera are collected through a video acquisition module and are transmitted to a video processing and analysis unit; a video processing module in the video processing and analysis unit pre-processes the data, including feature extraction, classification identification and image correction of the video image; then, the video analysis module and the computer vision module are used to identify the key points of the player's body in real time, such as shoulders, elbows, wrists, hips, knees and ankles, and track the movement trajectory thereof; the system can identify standard tennis actions, such as serving, forehand hitting, backhand hitting and intercepting, and compare the standard actions with a standard action model; through video analysis technology, the movement trajectory of the tennis ball is tracked in real time, and the flight speed, rotation and landing point of the ball are calculated; the system can identify different types of hitting, such as topspin, underspin and flat hitting, and analyze the effect thereof; through image recognition technology, the hitting point position of the player is accurately detected, and whether the timing and strength of the hitting are reasonable is analyzed; next, the video data and analysis results collected by the system are stored in a cloud database, so that subsequent query and comparative analysis are facilitated, the historical training data are supported to be traced back, and the training progress and effect are helped to be understood by the player and the coach; finally, through a real-time feedback module in a training feedback unit, the system can provide instant feedback to the player and the coach in the form of voice, text or image according to the real-time analysis results, for example, when the system detects that the serving action of the player is not standard, the player is immediately prompted to adjust the action; and based on a machine learning algorithm module and a deep learning algorithm module, the machine learning algorithm module comprises a decision tree algorithm, a random forest algorithm, a support vector machine algorithm and a neural network algorithm, the deep learning algorithm module comprises a convolutional neural network and a recurrent neural network, the system can generate suggestions and personalized training suggestions through a suggestion generation module and a personalized training formulation module according to the historical data and current performance of the player, for example, aiming at the weakness of the backhand hitting of the player, the system can recommend specific training plans and exercise actions.

[0014] Compared with the prior art, the application has the following beneficial effects:

[0015] Firstly, the multi-view high-definition camera and high-speed camera capture the training video of the athlete in real time, the video acquisition module collects the data of the multi-view high-definition camera and high-speed camera and transmits it to the video processing and analysis unit, the video processing module inside the video processing and analysis unit processes the data in advance, and then through the video analysis module and the computer vision module, the athlete's body key points are identified and tracked in real time, the system can identify the standard tennis action and compare it with the standard action model; through the video analysis technology, the motion trajectory of the tennis ball is tracked in real time, the system can identify different types of strokes and analyze their effects; through the image recognition technology, the athlete's hitting point position is accurately detected, and whether the timing and force of the stroke are reasonable is analyzed. The system stores the collected video data and analysis results in the cloud database, which is convenient for subsequent query and comparative analysis. Then, through the real-time feedback module inside the training feedback unit, the system can provide instant feedback to the athlete and the coach according to the real-time analysis results. Then, based on the machine learning algorithm module and the deep learning algorithm module, the system can generate suggestions and personalized training suggestions through the suggestion generation module and the personalized training module according to the athlete's historical data and current performance, in addition, the system also sets up a simulation training unit, which supports the integration of virtual reality equipment and augmented reality equipment.

[0016] In summary, the system can capture and analyze the athlete's action and ball trajectory in real time, provide instant feedback, and help the athlete adjust the action in the training process; through computer vision and machine learning technology, the system can accurately identify the athlete's action and hitting point, and provide high-precision data analysis; the system can generate personalized training suggestions according to the athlete's personal characteristics and training needs, and help the athlete improve the technical level; the system not only analyzes the athlete's action, but also analyzes the ball trajectory, hitting point and other dimensions, and provides comprehensive training evaluation; the system also supports integration with VR / AR devices, providing a highly simulated training environment to enhance the training effect. BRIEF DESCRIPTION OF DRAWINGS

[0017] Fig. 1 is a system block diagram of the present application;

[0018] Fig. 2 is a system block diagram of the video processing and analysis unit of the present application;

[0019] Fig. 3 is a system block diagram of the training suggestion and personalized training unit of the present application;

[0020] Fig. 4 is a system block diagram of the simulation training unit of the present application;

[0021] Fig. 5 is a structural schematic diagram of the video acquisition unit of the present application.

[0022] In the figure: 1, system cloud platform; 101, cloud database; 2, video acquisition unit; 3, video processing and analysis unit; 4, training feedback unit; 5, training suggestion and individualized training unit; 6, simulation training unit; 7, video acquisition module; 8, multi-view high-definition camera; 9, high-speed camera; 10, real-time feedback module; 11, video processing module; 12, video analysis module; 13, computer vision module; 14, machine learning algorithm module; 15, deep learning algorithm module; 16, suggestion generation module; 17, individualized training formulation module; 18, virtual reality device; 19, augmented reality device; 20, ring track; 21, motorized gimbal. DETAILED DESCRIPTION

[0023] In order to make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the claimed present application, but merely represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the scope of the present application.

[0024] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the scope of the present application.

[0025] Please refer to FIGS. 1-5, one embodiment provided by the present application: a tennis training auxiliary system based on tennis video, including system cloud platform 1, the inside of system cloud platform 1 includes video acquisition unit 2, video processing and analysis unit 3, training feedback unit 4, training suggestion and individualized training unit 5, simulation training unit 6, the inside of video acquisition unit 2 includes multi-view high-definition camera 8 and high-speed camera 9, the inside of video processing and analysis unit 3 includes video processing module 11, video analysis module 12, computer vision module 13;

[0026] Specifically, multi-view high-definition camera 8 and high-speed camera 9 capture the training video of the athlete in real time, multi-view high-definition camera 8 can be installed at different positions around the court to ensure full coverage of the athlete's movements and the movement trajectory of the ball, high-speed camera 9 is installed on the motorized gimbal 21 on the surface of the ring track 20, so that the high-speed camera 9 can do ring movement, which is convenient for multi-directional and dead-angle-free shooting and recording, and subsequent multi-dimensional analysis, the data of multi-view high-definition camera 8 and high-speed camera 9 are collected by video acquisition module 7 and transmitted to video processing and analysis unit 3;

[0027] Further, the video processing module 11 inside the video processing and analysis unit 3 processes the data in advance, including feature extraction, classification identification, image correction of video images, and then through the video analysis module 12 and the computer vision module 13, the system can identify the key points of the athlete's body in real time, such as shoulders, elbows, wrists, hips, knees, ankles, etc., and track their movement trajectories. The system can identify standard tennis actions such as serving, forehand, backhand, and volley, and compare them with standard action models. Through video analysis technology, the system can track the trajectory of the tennis ball in real time, calculate the speed, rotation, and landing point of the ball, and analyze the effect of different types of shots such as topspin, underspin, and flat shots. Through image recognition technology, the system can accurately detect the hitting point position of the athlete and analyze whether the timing and strength of the shot are reasonable. The system stores the collected video data and analysis results in the cloud database 101 for subsequent query and comparative analysis, supports the backtracking of historical training data, and helps athletes and coaches understand the training progress and effect;

[0028] The real-time feedback module 10 is inside the training feedback unit 4, and the training suggestion and individualized training unit 5 includes a machine learning algorithm module 14, a deep learning algorithm module 15, a suggestion generation module 16, and an individualized training development module 17;

[0029] Specifically, through the real-time feedback module 10 inside the training feedback unit 4, the system can provide immediate feedback to athletes and coaches in the form of voice, text, or images based on real-time analysis results. For example, when the system detects that the athlete's serving action is not standard, it will immediately prompt the athlete to adjust the action;

[0030] Further, based on the machine learning algorithm module 14 and the deep learning algorithm module 15, the machine learning algorithm module 14 includes decision tree algorithm, random forest algorithm, support vector machine algorithm, and neural network algorithm, and the deep learning algorithm module 15 includes convolutional neural network and recurrent neural network. The system can generate suggestions and individualized training suggestions through the suggestion generation module 16 and the individualized training development module 17 based on the athlete's historical data and current performance. For example, for the athlete's weakness in backhand shots, the system will recommend specific training plans and practice actions;

[0031] The system cloud platform 1 saves the system data in the cloud database 101 in real time through the wireless transmission module;

[0032] The multi-angle high-definition camera 8 and the high-speed camera 9 capture the training video of the athlete in real time, and the multi-angle high-definition camera 8 video acquisition unit 2 acquires the data of the multi-angle high-definition camera 8 and the high-speed camera 9 through the video acquisition module 7;

[0033] Multi-view high-definition camera 8 is installed at different positions around the court to ensure full coverage of the movement of the players and the trajectory of the ball, and high-speed camera 9 is installed on the electric pan-tilt head 21 on the surface of the annular track 20 in the court;

[0034] The video processing module 11 is used for feature extraction, classification identification and image correction of video images;

[0035] The video analysis module 12 performs action capture and identification, ball trajectory analysis and hitting point detection through the computer vision module 13;

[0036] The machine learning algorithm module 14 includes decision tree algorithm, random forest algorithm, support vector machine algorithm and neural network algorithm, and the deep learning algorithm module 15 includes convolutional neural network and recurrent neural network. The suggestion generation module 16 and the individualized training formulation module 17 generate training suggestion content and individualized training content based on the machine learning algorithm module 14 and the deep learning algorithm module 15 respectively;

[0037] The simulation training unit 6 includes virtual reality equipment 18 and augmented reality equipment 19;

[0038] In addition, the system is also provided with a simulation training unit 6, which supports the integration of virtual reality equipment 18 and augmented reality equipment 19. Athletes can conduct simulation training through virtual reality environment. The system can adjust the reaction of virtual opponents in real time according to the actions of athletes, providing a highly simulated training experience. In the augmented reality mode, the system can superimpose virtual ball trajectory and action guidance in the real scene, helping athletes better understand and correct their actions.

[0039] The embodiment of the application is used as follows: the system is composed of a video acquisition unit 2, a video processing and analysis unit 3, a training feedback unit 4, a training suggestion and individualized training unit 5, and a simulation training unit 6. First, a multi-angle high-definition camera 8 and a high-speed camera 9 capture the training video of the athlete in real time. The multi-angle high-definition camera 8 can be installed at different positions around the court to ensure that the athlete's movements and the trajectory of the ball are covered in all directions. The high-speed camera 9 is installed on a motorized pan-tilt head 21 on the surface of a ring-shaped track 20, so that the high-speed camera 9 can move in a ring shape, facilitating multi-directional and dead-angle-free shooting and recording, and facilitating subsequent multi-dimensional analysis. The data of the multi-angle high-definition camera 8 and the high-speed camera 9 are collected by a video acquisition module 7 and transmitted to the video processing and analysis unit 3. The video processing module 11 inside the video processing and analysis unit 3 processes the data in advance, including feature extraction, classification identification, and image correction of video images. Then, through the video analysis module 12 and the computer vision module 13, the athlete's body key points such as shoulders, elbows, wrists, hips, knees, and ankles are identified in real time, and their movement trajectories are tracked. The system can identify standard tennis actions such as serving, forehand, backhand, and volley, and compare them with standard action models. Through video analysis technology, the trajectory of the tennis ball is tracked in real time, and parameters such as the flight speed, rotation, and landing point of the ball are calculated. The system can identify different types of shots such as topspin, underspin, and flat shots, and analyze their effects. Through image recognition technology, the athlete's hitting point position is accurately detected, and the timing and strength of the shot are analyzed for reasonableness. The system stores the collected video data and analysis results in a cloud database 101 for subsequent query and comparative analysis, supports the backtracking of historical training data, and helps athletes and coaches understand the training progress and effects. Then, through the real-time feedback module 10 inside the training feedback unit 4, the system can provide immediate feedback to athletes and coaches in the form of voice, text, or images based on real-time analysis results. For example, when the system detects that the athlete's serving action is not standard, it will immediately prompt the athlete to adjust the action. Then, based on the machine learning algorithm module 14 and the deep learning algorithm module 15, the machine learning algorithm module 14 includes decision tree algorithm, random forest algorithm, support vector machine algorithm, and neural network algorithm, and the deep learning algorithm module 15 includes convolutional neural network and recurrent neural network. The system can generate suggestions and individualized training suggestions through the suggestion generation module 16 and the individualized training development module 17 based on the athlete's historical data and current performance. For example, for the athlete's weakness in backhand shots, the system will recommend specific training plans and practice actions. In addition, the system also has a simulation training unit 6 that supports the integration of virtual reality devices 18 and augmented reality devices 19. Athletes can conduct simulation training through virtual reality environments, and the system can adjust the reaction of virtual opponents in real time based on the athlete's actions, providing a highly simulated training experience.In the augmented reality mode, the system can superimpose virtual ball trajectory and action guidance in the real scene, helping the player better understand and correct the action.

[0040] In summary, the system can capture and analyze the player's action and ball trajectory in real time, provide immediate feedback, and help the player adjust the action in the training process; through computer vision and machine learning technology, the system can accurately identify the player's action and hitting point, providing high-precision data analysis; the system can generate personalized training recommendations according to the player's individual characteristics and training needs, helping the player improve their technical level; the system not only analyzes the player's action, but also analyzes the ball's trajectory, hitting point, and other dimensions, providing comprehensive training evaluation; the system also supports integration with VR / AR devices, providing a highly simulated training environment to enhance training effectiveness.

[0041] Obviously, the above-described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor should belong to the scope of protection of the present application.

Claims

1. A tennis training assistance system based on tennis video, characterized by, The system cloud platform (1) comprises a video acquisition unit (2), a video processing and analysis unit (3), a training feedback unit (4), a training suggestion and personalized training unit (5), and a simulation training unit (6). The video acquisition unit (2) comprises a multi-view high-definition camera (8) and a high-speed camera (9). The video processing and analysis unit (3) comprises a video processing module (11), a video analysis module (12), and a computer vision module (13). The training feedback unit (4) is provided with a real-time feedback module (10). The training suggestion and personalized training unit (5) comprises a machine learning algorithm module (14), a deep learning algorithm module (15), a suggestion generation module (16), and a personalized training formulation module (17).

2. The tennis training assistance system based on tennis video according to claim 1, characterized in that: The system cloud platform (1) saves system data in a cloud database (101) in real time through a wireless transmission module.

3. The tennis training assistance system based on tennis video according to claim 1, characterized in that: The multi-view high-definition camera (8) and the high-speed camera (9) capture training videos of athletes in real time. The video acquisition unit (2) collects data of the multi-view high-definition camera (8) and the high-speed camera (9) through a video acquisition module (7).

4. The tennis training assistance system based on tennis video according to claim 1, characterized in that: The multi-view high-definition camera (8) is installed at different positions around the court to ensure full coverage of the movements of athletes and the movement trajectory of the ball. The high-speed camera (9) is installed on a motorized gimbal (21) on the surface of a ring-shaped track (20) in the court.

5. The tennis training assistance system based on tennis video according to claim 1, characterized in that: The video processing module (11) is used for feature extraction, classification identification, and image correction of video images.

6. The tennis training assistance system based on tennis video according to claim 1, characterized in that: The video analysis module (12) performs action capture and identification, ball trajectory analysis, and hitting point detection through the computer vision module (13).

7. The tennis training assistance system based on tennis video according to claim 1, characterized in that: The machine learning algorithm module (14) comprises decision tree algorithm, random forest algorithm, support vector machine algorithm, and neural network algorithm. The deep learning algorithm module (15) comprises convolutional neural network and recurrent neural network. The suggestion generation module (16) and the personalized training formulation module (17) generate training suggestion content and personalized training content based on the machine learning algorithm module (14) and the deep learning algorithm module (15), respectively.

8. The tennis training assistance system based on tennis video according to claim 1, characterized in that: The simulation training unit (6) comprises a virtual reality device (18) and an augmented reality device (19).

9. A tennis training assistance method based on a tennis video, characterized by: Firstly, the training video of the player is captured in real time by the multi-view high-definition camera (8) and the high-speed camera (9), the multi-view high-definition camera (8) can be installed at different positions around the court to ensure full coverage of the player's movements and the trajectory of the ball, the high-speed camera (9) is installed on the electric pan-tilt (21) on the surface of the ring track (20), so that the high-speed camera (9) can make ring movement, which is convenient for multi-directional and dead-angle-free shooting and recording, and convenient for subsequent multi-dimensional analysis; Then, the data of the multi-view high-definition camera (8) and the high-speed camera (9) are collected by the video acquisition module (7) and transmitted to the video processing and analysis unit (3), the video processing module (11) inside the video processing and analysis unit (3) processes the data in advance, including feature extraction, classification identification and image correction of video images, and then through the video analysis module (12) and the computer vision module (13), the body key points of the player are identified in real time, such as shoulders, elbows, wrists, hips, knees and ankles, and the movement trajectory is tracked, the system can identify standard tennis actions, such as serving, forehand hitting, backhand hitting and intercepting, and compare with the standard action model; Through video analysis technology, the movement trajectory of the tennis ball is tracked in real time, and the parameters such as the flight speed, rotation and landing point of the ball are calculated, the system can identify different types of hitting, such as topspin, underspin and flat hitting, and analyze the effect; Through image recognition technology, the hitting point position of the player is accurately detected, and whether the timing and strength of the hitting are reasonable is analyzed; Next, the system stores the collected video data and analysis results in the cloud database (101), which is convenient for subsequent query and comparative analysis, supports the backtracking of historical training data, and helps the player and the coach to understand the training progress and effect; Finally, through the real-time feedback module (10) inside the training feedback unit (4), the system can provide immediate feedback to the player and the coach in the form of voice, text or image according to the real-time analysis results, for example, when the system detects that the player's serving action is not standard, the player will be prompted to adjust the action immediately, and based on the machine learning algorithm module (14) and the deep learning algorithm module (15), the machine learning algorithm module (14) includes decision tree algorithm, random forest algorithm, support vector machine algorithm and neural network algorithm, the deep learning algorithm module (15) includes convolutional neural network and recurrent neural network, the system can generate suggestions and personalized training suggestions through the suggestion generation module (16) and the personalized training formulation module (17) according to the historical data and current performance of the player, for example, for the weakness of the player's backhand hitting, the system will recommend specific training plan and exercise action.

Images