Intelligent training method and system for realizing full technical action analysis of tennis

Abstract

The application provides an intelligent training method and system for realizing full technical action analysis of tennis, which comprises the following steps: acquiring a training video stream of a trainer performing tennis action training in a training field in real time; extracting action motion features from video image frames of the training video stream, and identifying an action type of a current tennis action according to the action motion features; acquiring a target biomechanics analysis model matched with the action type from a preset model library according to the action type; calling the target biomechanics analysis model to perform biomechanics analysis on the current tennis action to determine key action parameters of the current tennis action; comparing the key action parameters with standard action parameters corresponding to the current tennis action in a preset expert action database, combining a comparison result and the key action parameters to perform problem diagnosis on the current tennis action training of the trainer and generate improvement suggestions; and feeding back the problem diagnosis result and the improvement suggestions of a single tennis action in real time.

Description

Technical Field

[0001] This invention relates to the field of sports training and analysis technology, and in particular to an intelligent training method and system for realizing the analysis of all tennis technical movements. Background Technology

[0002] Tennis is a comprehensive sport involving a variety of complex techniques. An excellent player not only needs to master high-quality serves but also must possess consistent baseline forehand and backhand strokes, quick volleys at the net, and agile footwork on the court. Therefore, comprehensive and efficient training of all techniques is key to improving a player's skill level. Currently, while many products and solutions on the market attempt to assist tennis training through technology, they generally suffer from deeper problems such as limited functionality, fragmented scenarios, and insufficient intelligence, failing to provide a one-stop, comprehensive guidance.

[0003] Current tennis training programs and their technical status mainly include the following categories: The first category: specialized, standalone smart hardware. These products typically focus on measuring or analyzing a specific technical aspect or data dimension, creating functional "data silos." Examples include wearable / integrated sensors like the Babolat Play smart racket or the Zepp Labs tennis swing analyzer. Through built-in gyroscopes and accelerometers, they can accurately measure swing speed, the distribution of the impact point on the racket face, and the type of ball spin (topspin / flat / slice). However, these products have limited data dimensions. These devices are essentially "extensions of the arm and racket," and their data collection is limited to this. They cannot address the root causes that determine shot quality—such as lower limb push-off, core rotation and energy storage, and the sequence of the body's kinetic chain. It can tell you "swing speed is slow," but it cannot tell you "why it's slow," let alone diagnose whether it's due to "insufficient hip rotation" or "incorrect push-off timing." Furthermore, they lack morphological feedback. They provide abstract numbers rather than intuitive feedback on movement patterns. Trainees cannot determine from this whether their ready posture is correct, whether their backswing trajectory is standard, or whether there are any problems with their body balance.

[0004] The second category consists of single-function automated training equipment. This is represented by various automatic tennis ball machines, such as products from brands like Lobster Sports or Spinshot. These devices solve the problem of "someone feeding the ball," allowing for the setting of different landing points, frequencies, and spins for high-intensity, repetitive hitting practice. However, these products provide the "feeding" but lack "teaching," lacking an intelligent feedback loop. The ball machine is merely a mechanical, one-way "ball-feeding tool." It doesn't know if the trainee's movements are correct, nor can it evaluate the quality of the hits. When using the ball machine, the trainee is essentially still in a state of "blind repetition," and without a coach's guidance, incorrect movements can become ingrained through extensive practice. It solves the physical problem of multi-ball practice, but it doesn't address the cognitive problem of technique improvement.

[0005] The third category: "Smart Court" solutions based on fixed cameras. Represented by systems like Playsight or Wingfield, these systems use multiple cameras permanently installed around the court to automatically record the entire match, compile data (such as scores, landing point distribution, first serve success rate), and edit highlight reels. However, these products focus on "post-match review" rather than "real-time guidance," resulting in significant feedback delays. The core function of these systems is "match recording and analysis," and users typically need to wait until the entire match or training session is over before viewing data and replays via an app or courtside terminal. This delay, measured in hours, fails to meet the needs of trainees for immediate adjustments during training. Furthermore, these products lack in-depth biomechanical analysis. Their algorithms primarily focus on ball trajectory tracking and event detection (such as out-of-bounds and points scored), offering relatively superficial analysis of the trainee's "technical movements." They might tell you "the ball went into the net," but usually don't delve into the biomechanical reasons that led to it. High cost and complete immobility are also significant drawbacks. "Smart courts" involve a massive "asset" transformation of the entire court, with high installation costs, and once installed, they cannot be moved. This makes the program only available to a few high-end clubs and unsuitable for individual users, coaches, or schools training in different venues. Summary of the Invention

[0006] In view of the above problems, the present invention is proposed to provide an intelligent training method and system for realizing the analysis of all tennis technical movements, which solves or at least partially solves the above technical problems.

[0007] One aspect of the present invention provides an intelligent training method for realizing full technical motion analysis in tennis, the method comprising: Acquire real-time training video streams of trainees practicing tennis moves in the training venue; Extract motion features from video image frames of the training video stream, and identify the motion type of the current tennis move based on the motion features; Based on the action type, a target biomechanical analysis model matching the action type is obtained from a preset model library; The target biomechanical analysis model is invoked to perform biomechanical analysis on the current tennis motion to determine the key motion parameters of the current tennis motion. The key motion parameters are compared with the standard motion parameters corresponding to the current tennis motion in the preset expert motion database. Based on the comparison results and the key motion parameters, the trainee's current tennis motion training is diagnosed and improvement suggestions are generated. The diagnostic results and improvement suggestions for a single tennis move will be provided in real time.

[0008] Furthermore, after providing real-time feedback on the diagnostic results and improvement suggestions for a single tennis move, the method also includes: Key motion parameters, problem diagnosis results, and improvement suggestions for a single tennis move are stored in the current trainee's personal ability profile database. Based on the key motion parameters, problem diagnosis results, and improvement suggestions for each tennis move collected from the individual ability profile database, a comprehensive ability radar chart and key motion parameter change curves are generated.

[0009] Furthermore, the method also includes: Based on the identified movement type and / or problem diagnosis results, target guidance parameters matching the current tennis movement training are obtained, and laser guidance routes or areas are projected according to the target guidance parameters to guide the trainee to follow the guidance for the next round of training of the current tennis movement.

[0010] Furthermore, the acquisition of target guidance parameters matching the current tennis movement training includes: Obtain key physiological data from trainees; The laser guidance parameters of the guidance device are calibrated based on the key physiological data, and target guidance parameters that match the trainee's current tennis movement training are extracted from the laser guidance parameters.

[0011] Furthermore, when the tennis action type is a serve, the step of calling the target biomechanical analysis model to perform biomechanical analysis on the current tennis action to determine the key action parameters of the current tennis action includes: The key joints of the human body in the video image frames corresponding to the serving action are identified to extract three-dimensional skeleton points, and a human motion dynamic model is constructed based on the temporal changes of each three-dimensional skeleton point. Based on the extraction results of each three-dimensional skeletal point, the hip-shoulder separation angle formed by hip rotation and shoulder rotation during the movement cycle is calculated, and the trunk twisting amplitude and the body's power transmission efficiency are determined according to the hip-shoulder separation angle. Calculate the angle changes and angular velocity timing of each three-dimensional skeletal point during the movement process. Analyze the order of force exertion and force transmission efficiency of each joint from the lower limbs and trunk to the upper limbs based on the angle changes and angular velocity timing. Determine whether the order of force exertion of the limbs is correct based on the order of force exertion and force transmission efficiency of each joint. Identify the tennis ball in the video image frame corresponding to the serve action, fit the three-dimensional motion trajectory of the ball based on the continuous motion coordinates of the tennis ball, compare the three-dimensional motion trajectory of the ball with the preset standard trajectory, and obtain the trajectory spatial offset and the spatial error of the highest point of the ball throw. The system identifies the three-dimensional spatial coordinates of the tennis ball and the hitting tool at the moment of impact, locates the actual hitting position, and compares the actual hitting position with the standard ideal hitting point to calculate the deviation of the hitting point position.

[0012] Furthermore, when the tennis motion type is a forehand, the step of calling the target biomechanical analysis model to perform biomechanical analysis on the current tennis motion to determine the key motion parameters of the current tennis motion includes: Identify the time frame of the single swing backwards limit in the video image frame corresponding to the forehand action. In the current time frame, with the trainee's torso center axis as the spatial reference plane, extract the horizontal offset distance and vertical downward sinking distance of the racket head relative to the spatial reference plane. Calculate the combined vector length through a spatial vector synthesis algorithm to obtain the maximum backwards amplitude. Acquire the vertical displacement time-series data and three-axis angular velocity time-series data of the racket head throughout the entire swing process. Identify the timestamp T1 of the lowest point of the swing and the timestamp T2 of the contact point in the corresponding video image frame of the forehand action. Calculate the average value of the instantaneous vertical acceleration of the racket head in the interval from T1 to T2 using differential calculation based on the vertical displacement time-series data and the three-axis angular velocity time-series data, and obtain the vertical acceleration of the racket head from the lowest point to the contact point.

[0013] Another aspect of the present invention provides an intelligent training system for realizing full technical motion analysis in tennis, the system comprising a physical structure body and a vision and environment acquisition unit, a motion execution and interaction unit, and a platform-based AI processing center deployed on the physical structure body; The vision and environment acquisition unit is used to acquire training video streams of trainees practicing tennis moves in the training venue in a non-contact manner in real time and transmit them to the platform-based AI processing center. A platform-based AI processing center is used to execute the intelligent training method for realizing the analysis of all technical movements in tennis as described in any one of claims 1-6; The motion execution and interaction unit is used to visualize the problem diagnosis results and improvement suggestions of a single tennis move, which are fed back in real time by the platform-based AI processing center.

[0014] Furthermore, the vision and environment acquisition unit includes at least one 3D depth camera and at least one high-speed camera, or a combination of a 3D depth camera, a high-speed camera, and a wide-angle camera, for training video streaming when a trainee performs tennis action training in the training venue.

[0015] Furthermore, the system also includes: The laser projection unit is used to project laser guidance paths or areas onto the training area to guide the trainee in following the guidance for the next round of training of the current tennis movement.

[0016] Furthermore, the system also includes a dedicated ball supply unit for serve training; The dedicated ball supply unit for serve training includes: The storage and gravity ball supply module includes a ball storage compartment and a ball delivery slide, which is used to transport the tennis balls in the ball storage compartment to the release buffer through the ball delivery slide using the principle of gravity; A buffer to be released, used to load and position tennis balls for free-fall training from the ball release chute; A tennis ball release device is used to control the release of a tennis ball from a buffer to be released.

[0017] The intelligent training method and system for analyzing all tennis techniques provided in this invention can automatically identify the type of tennis movements in real-time training. It uses a biomechanical analysis model matching the movement type to perform biomechanical analysis on the current tennis movement to determine key movement parameters. These parameters are then compared with standard movement parameters corresponding to the current tennis movement in a preset expert movement database. The comparison results and key movement parameters are combined to diagnose problems in the trainee's current tennis movement training, and the diagnostic results and improvement suggestions are provided in real-time. This represents a leap from isolated functionalities to a unified platform, providing unprecedented holistic and coherent training, ensuring more comprehensive and accurate training analysis, and achieving efficient skill improvement. This invention can be used not only for structured multi-ball practice but also for near-real-game two-person rallies or high-intensity multi-ball training, analyzing every non-preset hitting motion in real time. This greatly expands the application scenarios of intelligent training equipment, upgrading it from a "practice tool" to a "competitive partner."

[0018] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0019] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. In the drawings: Figure 1 This is a structural block diagram of an intelligent training system for analyzing all tennis techniques, according to an embodiment of the present invention. Figure 2 This is a structural block diagram of an intelligent training system for realizing full technical motion analysis in tennis, according to another embodiment of the present invention. Figure 3 This is a flowchart illustrating an intelligent training method for analyzing all tennis techniques, as described in an embodiment of the present invention. Detailed Implementation

[0020] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0021] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art and should not be interpreted in an idealized or overly formal sense unless specifically defined.

[0022] To overcome the shortcomings of existing technologies, such as limited equipment functionality, fragmented training scenarios, and insufficient intelligence, this invention provides an intelligent training system that enables comprehensive analysis of all tennis techniques. Within a unified system, this invention utilizes an AI engine capable of automatically recognizing various technical movements to achieve comprehensive analysis and management of all core tennis skills. Employing a flexible and quickly deployable hardware structure (such as an L-shaped cantilever), the entire professional analysis system can be easily moved and deployed across different courts, allowing professional guidance to "move with the user." This is extremely user-friendly and conducive to the training environment, maintaining real-time feedback capabilities unmatched by "smart courts," while providing comprehensive biomechanical diagnostics far exceeding those of ordinary sensors and apps, achieving the best of both worlds.

[0023] like Figure 1 As shown, the intelligent training system for realizing full technical motion analysis in tennis proposed in this embodiment includes a physical structure ontology and a vision and environment acquisition unit 100, a motion execution and interaction unit 200, and a platform-based AI processing center 300 deployed on the physical structure ontology, wherein: The vision and environment acquisition unit 100 is used to acquire training video streams of trainees practicing tennis movements in the training venue in a non-contact manner in real time and transmit them to the platform-based AI processing center. The platform-based AI processing center 300 is used to execute the intelligent training method for realizing the analysis of all technical movements in tennis as described in any one of claims 1-6; The motion execution and interaction unit 200 is used to visualize the problem diagnosis results and improvement suggestions of a single tennis movement fed back in real time by the platform-based AI processing center.

[0024] In one specific embodiment of the present invention, the physical structure body preferably adopts a portable cantilever (L-shaped) structure to adapt to flexible deployment in different positions on the court (such as the baseline, service area, and net area).

[0025] In one specific embodiment of the present invention, such as Figure 2 As shown, the vision and environment acquisition unit 100 includes at least one 3D depth camera 110 and at least one high-speed camera 120, or a combination of the 3D depth camera 100, the high-speed camera 120, and the wide-angle camera 130, for use as a training video stream when a trainee performs tennis movements on a training court. In this embodiment, the vision and environment acquisition unit 100 is used to capture any technical movements performed by the trainee at any location on the court. To this end, a wide-angle camera (130) can be added to cover a wider area, ensuring that movements can be fully captured even in dynamic scenes such as rallies. The vision and environment acquisition unit 100 provides the platform-based AI processing center 300 with a high-quality, wide-scene image data stream input.

[0026] In this embodiment, as Figure 2 As shown, the platform-based AI processing center 300 is the core of the intelligent training system, designed to address the problems of fragmented functions and insufficient intelligence in existing technologies. It mainly includes: The automatic technical movement recognition engine 310 receives training video streams from the acquisition unit 100 and, through a pre-trained deep learning classification model, automatically and in real time determines the type of technical movement the trainee is currently performing. For example, it outputs labels such as "forehand open stance topspin" or "backhand double backhand straight shot," addressing the key issue that existing techniques require manual mode selection and cannot adapt to real-world scenarios. It dynamically invokes the most suitable biomechanical analysis model for specialized in-depth analysis, diagnoses problems in the trainee's current tennis movement training based on the analysis results, generates improvement suggestions, and provides real-time feedback.

[0027] The specialized biomechanical analysis model library 320 is an extensible collection of models, pre-configured with an independent, deeply optimized biomechanical analysis model for each important tennis technique, such as serve, forehand, backhand, and volley. When the automatic technique recognition engine 310 outputs the "forehand" label, the system automatically calls the forehand analysis model (Model B); when "serve" is recognized, the serve analysis model (Model A) is called. Each model focuses on calculating the most critical biomechanical parameters of the corresponding movement.

[0028] Expert Movement Database 340: Contains standard data for all technical movements from professional athletes.

[0029] The Personal Capability Profile Generation Module 330 enables the analysis data of various technical actions to no longer be stored separately like "data silos," but to aggregate the analysis results of all different technical actions, creating a unified, multi-dimensional "personal capability profile" for users, such as displaying it in the form of a capability radar chart, and tracking the long-term progress curves of various skills.

[0030] Furthermore, the platform-based AI processing center 300 is specifically used to acquire real-time training video streams of trainees practicing tennis movements in the training venue; extract motion features from the video image frames of the training video stream, and identify the movement type of the current tennis movement based on the motion features; acquire a target biomechanical analysis model matching the movement type from a preset model library based on the movement type; call the target biomechanical analysis model to perform biomechanical analysis on the current tennis movement to determine the key movement parameters of the current tennis movement; compare the key movement parameters with the standard movement parameters corresponding to the current tennis movement in a preset expert movement database, and combine the comparison results and the key movement parameters to diagnose problems in the trainee's current tennis movement training and generate improvement suggestions; and provide real-time feedback on the problem diagnosis results and improvement suggestions for a single tennis movement.

[0031] It's worth noting that, in addition to the two-step "recognition + invocation" architecture, a larger and more complex single deep learning model can be trained. This model takes the raw video stream as input and outputs specific biomechanical parameters and diagnostic results for that action. For example, the neural network layers within the model automatically learn to distinguish different actions and activate corresponding analysis paths. Alternatively, the trainee's continuous skeletal joint data can be treated as a time series. Using a Transformer model similar to those in natural language processing, its self-attention mechanism can not only identify the type of the current action but also capture the correlation between preceding and subsequent actions. For instance, it can analyze the trainee's physical decline and movement deformation trends during a series of consecutive multi-ball exercises.

[0032] Furthermore, after providing real-time feedback on the problem diagnosis results and improvement suggestions for a single tennis move, the key movement parameters, problem diagnosis results, and improvement suggestions for that single tennis move are stored in the current trainee's personal ability profile database. Based on the key movement parameters, problem diagnosis results, and improvement suggestions for each tennis move collected in the personal ability profile database, a comprehensive ability radar chart and key movement parameter change curves are generated.

[0033] Furthermore, the platform-based AI processing center 300 is specifically used to obtain target guidance parameters matching the current tennis movement training based on the identified movement type and / or problem diagnosis results, and to project a laser guidance route or area according to the target guidance parameters to guide the trainee to follow the guidance for the next round of training of the current tennis movement. Obtaining the target guidance parameters matching the current tennis movement training includes: acquiring the trainee's key physiological data; calibrating the laser guidance parameters of the guidance device based on the key physiological data; and extracting the target guidance parameters matching the trainee's current tennis movement training from the laser guidance parameters.

[0034] In an optional embodiment, when the tennis action is a serve, the step of calling the target biomechanical analysis model to perform biomechanical analysis on the current tennis action to determine the key action parameters of the current tennis action includes: identifying key human joints in the video image frame corresponding to the serve action to extract three-dimensional skeletal points; constructing a human motion dynamic model based on the temporal changes of each three-dimensional skeletal point; calculating the hip-shoulder separation angle formed by hip rotation and shoulder rotation during the action cycle based on the extraction results of each three-dimensional skeletal point; and determining whether the trunk twisting amplitude and the body's power transmission efficiency meet the standards based on the hip-shoulder separation angle; and calculating the angle changes and angular velocities of each three-dimensional skeletal point during the movement. The timing analysis, based on angle changes and angular velocity timing, examines the sequence of force exertion and power transmission efficiency of each joint from the lower limbs, trunk, to the upper limbs, and determines the correctness of the force exertion sequence based on the sequence of force exertion and power transmission efficiency. It identifies the tennis ball in the corresponding video frame of the serving action, fits the ball's three-dimensional trajectory based on its continuous motion coordinates, compares the three-dimensional trajectory with a preset standard trajectory to obtain the trajectory spatial offset and the spatial error of the highest point of the ball's toss. Finally, it identifies the three-dimensional spatial coordinates of the tennis ball and the hitting tool at the moment of impact, locates the actual hitting spatial position, and compares the actual hitting spatial position with the standard ideal hitting point to calculate the hitting point position deviation.

[0035] In an optional embodiment, when the tennis action type is a forehand, the step of calling the target biomechanical analysis model to perform biomechanical analysis on the current tennis action to determine the key action parameters of the current tennis action includes: identifying the time sequence frame of the single backswing limit critical point in the video image frame corresponding to the forehand action; taking the trainee's torso center axis as the spatial reference plane in the current time sequence frame, extracting the horizontal offset distance and vertical downward sinking distance of the racket head relative to the spatial reference plane, respectively, and calculating the resultant vector length through a spatial vector synthesis algorithm to obtain the maximum backswing amplitude; acquiring the racket head vertical displacement time sequence data and triaxial angular velocity time sequence data throughout the entire swing process; identifying the racket lowest point timestamp T1 and the ball contact point timestamp T2 in the video image frame corresponding to the forehand action; and using differential operation to calculate the average value of the racket head vertical instantaneous acceleration in the interval from T1 to T2 based on the vertical displacement time sequence data and triaxial angular velocity time sequence data to obtain the racket head vertical acceleration from the lowest point to the point of contact.

[0036] In this embodiment, the motion execution and interaction unit 200 is the window through which the trainee 210 exchanges information with the system. Its core is the display and operation terminal 220, which can not only display the instant feedback of a single action, but also show the comprehensive ability profile and long-term progress curve generated by the module 330, so that the user can have a macro-level understanding of his overall technical level.

[0037] Furthermore, the intelligent training system for analyzing all tennis techniques provided in this embodiment of the invention also includes a laser projection unit, used to project laser guidance routes or areas onto the training area to guide the trainee in the next round of training for the current tennis movement. In this embodiment, the laser projection unit can dynamically project recommended footwork movement routes for forehand and backhand shots or optimal preparation areas for volleys, etc., based on the movements identified and problems diagnosed by the platform-based AI processing center 300, to guide the trainee in training.

[0038] Furthermore, the intelligent training system for realizing full-technical motion analysis in tennis provided by this embodiment of the invention also includes a dedicated ball supply unit for serve training. This dedicated ball supply unit includes a storage and gravity-feed module, a release buffer, and a tennis ball release device. The various parts of the dedicated ball supply unit can be packaged into a "pluggable" module. When a user needs to perform specific serve training, this module can be quickly installed; when performing baseline rally analysis or other training, it can be removed, making the system lighter and more flexible. Specifically, the storage and gravity-feed module includes a ball storage compartment and a ball release chute, used to transport tennis balls from the storage compartment to the release buffer via the ball release chute using gravity. The release buffer is used to load and position tennis balls from the release chute for free-fall training. The tennis ball release device is used to control the release of the tennis balls from the release buffer.

[0039] This invention employs a modular design, enabling a core system (acquisition unit + AI center + interaction unit) to flexibly adapt to various training scenarios, from single-person decomposition practice to two-person sparring, by adding or removing different peripheral modules. This completely solves the problems of fixed functions and limited scenarios in existing equipment. Furthermore, the preferred portable L-shaped structure ensures that the entire system (even including optional modules) can be easily moved and deployed by 1-2 people, requiring almost no modification to the training site, making it extremely user-friendly.

[0040] The core algorithm and hardware architecture of this invention are universal and can be directly applied to badminton (analyzing the whipping motion of a smash), table tennis (analyzing the fine wrist movements of the ball on the table), golf (analyzing the swing plane and body rotation), baseball (analyzing the swing motion), etc.

[0041] The trainee (210) practices hitting the ball at will on the field. The vision and environment acquisition unit 100 captures the movements and transmits them to the platform-based AI processing center 300. The platform-based AI processing center 300 first uses the automatic action recognition engine 310 to "understand" what the movement is, then retrieves the most suitable "expert" from the model library 320 to analyze it, and sends the analysis results and diagnostic suggestions to the motion execution and interaction unit 200 for real-time feedback to the trainee, while updating the trainee's personal ability profile. If needed, the platform-based AI processing center 300 can also control the laser projection unit to provide laser guidance to assist the trainee in the next round of training.

[0042] The intelligent training system for analyzing all tennis techniques provided by this invention has the following beneficial effects: First, it represents a leap from "functional silos" to a "unified platform," providing unprecedented holistic and coherent training. Existing technologies suffer from fragmented functionality, with data on serves, swings, footwork, etc., scattered across different devices or apps. This invention, through a unified platform, seamlessly analyzes and manages all of a trainee's technical movements. It eliminates data silos: data from all technical aspects is aggregated into a personal ability profile generation module. This enables the system to perform cross-technical correlation analysis; for example, it can reveal the intrinsic link between "low ball toss stability (serve data) and poor core strength (forehand and backhand data)," providing deeper, more comprehensive technical insights that no independent device can achieve. It provides a seamless training experience: users can view all training data, progress curves, and improvement suggestions in a unified interface without switching between different devices, resulting in a smooth and efficient training experience.

[0043] Secondly, by introducing an "automatic motion recognition engine," a leap from "manual" to "truly intelligent" has been achieved, greatly expanding application scenarios. Most existing systems require users to manually select training modes, making them unsuitable for real and varied training environments. This invention completely solves this bottleneck through AI automatic recognition technology. It liberates the user and adapts to real-world scenarios: the system can automatically identify whether the user is performing a forehand shot, backhand slice, or net volley, without any manual intervention. This allows the invention not only for structured multi-ball practice but also, for the first time, for application in near-real-game two-person rallies or high-intensity multi-ball training, analyzing every non-preset shot in real time. This greatly broadens the application boundaries of intelligent training equipment, upgrading it from a "practice tool" to a "competition partner." It ensures immersive training: users can fully immerse themselves in training, forming a focused "flow" state without being interrupted by trivial matters such as operating the equipment, thereby improving the quality and enjoyment of training.

[0044] Third, adopting a modular and portable design concept, it combines the depth of professional analysis with the ultimate flexibility of scenario deployment. Existing technologies often fall into the dilemma of "functionality" and "portability" being mutually exclusive: powerful "smart courts" cannot be moved, while portable sensors have too limited functionality. This invention solves this contradiction through innovative design. High scenario adaptability: The optimized portable L-shaped structure allows the entire professional analysis system to be easily transported and quickly deployed at any location (baseline, net) on any tennis court. This allows professional guidance to "move with the person," greatly serving individual users, itinerant coaches, and schools with limited facilities. On-demand modular functionality: Modules such as the dedicated ball supply module are designed as "optional plug-ins," allowing the core system to remain extremely lightweight when these functions are not needed (such as during match analysis). This modular approach of "core system + optional peripherals" ensures the comprehensiveness of the platform's functions while optimizing hardware costs and deployment complexity, offering far better value than purchasing multiple independent devices or modifying the entire court.

[0045] Fourth, it provides a higher-level analysis, moving from "single-time diagnosis" to "long-term ability profile," enabling scientific management of trainee growth. Most existing technologies remain at the level of judging the right or wrong of a single action. This invention, through a personal ability profile generation module, elevates guidance from the "micro" to the "macro" level. It provides a global perspective: through a comprehensive ability radar chart, users can clearly see their strengths and weaknesses (e.g., excellent strength but insufficient stability), thus enabling more strategic training planning. It quantifies the growth trajectory: the long-term progress curve visualizes and quantifies the trainee's efforts and progress, which is invaluable for maintaining training motivation and scientifically adjusting training plans.

[0046] Figure 3 The flowchart illustrates an intelligent training method for analyzing all tennis techniques, as provided in an embodiment of the present invention. Figure 3 As shown, the intelligent training method for realizing full technical motion analysis in tennis proposed in this invention includes the following steps: S1. Acquire real-time training video streams of trainees practicing tennis moves in the training venue. S2. Extract motion features from video image frames of the training video stream, and identify the motion type of the current tennis move based on the motion features; S3. Obtain a target biomechanical analysis model that matches the action type from a preset model library according to the action type; S4. Call the target biomechanical analysis model to perform biomechanical analysis on the current tennis motion to determine the key motion parameters of the current tennis motion; S5. Compare the key motion parameters with the standard motion parameters corresponding to the current tennis motion in the preset expert motion database, and combine the comparison results with the key motion parameters to diagnose problems in the trainee's current tennis motion training and generate improvement suggestions. S6. Provide real-time feedback on the diagnostic results and improvement suggestions for a single tennis move.

[0047] In this embodiment of the invention, the system is first initialized and deployed. The user will use a portable core system with an L-shaped structure to quickly deploy it at any suitable location next to the training ground within minutes. After the system starts up, it automatically completes environmental calibration, fully demonstrating the system's portability and site adaptability.

[0048] Trainees can perform any form of training without any prior interaction with the device. The system silently and continuously captures the training video stream as the trainee practices tennis movements on the training court in the background. This invention provides a distraction-free, immersive training experience, allowing users to focus entirely on the tennis ball itself, rather than operating the device.

[0049] For each action captured from video image frames in the training video stream, motion features are extracted based on a pre-trained AI model. The motion type of the current tennis action is identified based on the motion features. A target biomechanical analysis model matching the current action type is obtained from a pre-set model library based on the action type. The target biomechanical analysis model is then called to perform biomechanical analysis on the current tennis action to determine the key motion parameters of the current tennis action.

[0050] This invention achieves true intelligence and context awareness, enabling it to "understand" the actions taking place on the court and analyze them using the most professional "knowledge." Furthermore, in addition to the parameters for serving, this invention also introduces specific parameters for other actions. For example: For topspin forehand shots, the focus will be on analyzing the "maximum backswing amplitude" and the "vertical acceleration of the racket head from the lowest point to the point of impact" to assess the power generation and the quality of the "wiper-like" swing.

[0051] For net volleys, the focus will be on analyzing the "reaction time (in milliseconds)" from the moment the opponent hits the ball to the moment we touch it, and the "racket face opening angle" at the moment of impact, in order to assess reaction speed and ball control feel.

[0052] In one specific embodiment, when the tennis action is a serve, the step of calling the target biomechanical analysis model to perform biomechanical analysis on the current tennis action to determine the key action parameters of the current tennis action includes: identifying key human joints in the video image frame corresponding to the serve action to extract three-dimensional skeletal points; constructing a human motion dynamic model based on the temporal changes of each three-dimensional skeletal point; calculating the hip-shoulder separation angle formed by hip rotation and shoulder rotation during the action cycle based on the extraction results of each three-dimensional skeletal point; and determining whether the trunk twisting amplitude and the body's power transmission efficiency meet the standards based on the hip-shoulder separation angle; and calculating the angle changes and angular velocities of each three-dimensional skeletal point during the movement. The timing analysis, based on angle changes and angular velocity timing, examines the sequence of force exertion and power transmission efficiency of each joint from the lower limbs, trunk, to the upper limbs, and determines the correctness of the force exertion sequence based on the sequence of force exertion and power transmission efficiency. It identifies the tennis ball in the corresponding video frame of the serving action, fits the ball's three-dimensional trajectory based on its continuous motion coordinates, compares the three-dimensional trajectory with a preset standard trajectory to obtain the trajectory spatial offset and the spatial error of the highest point of the ball's toss. Finally, it identifies the three-dimensional spatial coordinates of the tennis ball and the hitting tool at the moment of impact, locates the actual hitting spatial position, and compares the actual hitting spatial position with the standard ideal hitting point to calculate the hitting point position deviation.

[0053] In one specific embodiment, when the tennis action type is a forehand, the step of calling the target biomechanical analysis model to perform biomechanical analysis on the current tennis action to determine the key action parameters of the current tennis action includes: identifying the time sequence frame of the single swing backswing limit critical point in the video image frame corresponding to the forehand action; taking the trainee's torso center axis as the spatial reference plane in the current time sequence frame, extracting the horizontal offset distance and vertical downward sinking distance of the racket head relative to the spatial reference plane, respectively, and calculating the resultant vector length through a spatial vector synthesis algorithm to obtain the maximum backswing amplitude; acquiring the racket head vertical displacement time sequence data and triaxial angular velocity time sequence data throughout the swing process; identifying the racket lowest point timestamp T1 and the ball contact point timestamp T2 in the video image frame corresponding to the forehand action; and using differential operation to calculate the average value of the racket head vertical instantaneous acceleration in the interval from T1 to T2 based on the vertical displacement time sequence data and triaxial angular velocity time sequence data to obtain the racket head vertical acceleration from the lowest point to the point of contact.

[0054] Furthermore, after providing real-time feedback on the problem diagnosis results and improvement suggestions for a single tennis move, the method also includes: storing the key movement parameters, problem diagnosis results, and improvement suggestions for a single tennis move into the current trainee's personal ability profile database; and generating a comprehensive ability radar chart and key movement parameter change curves based on the key movement parameters, problem diagnosis results, and improvement suggestions for each tennis move collected in the personal ability profile database.

[0055] This invention provides dual feedback, offering both immediate feedback and comprehensive profile updates. On one hand, it provides immediate and specific diagnosis of the recently completed movement; on the other hand, it integrates this data into the user's long-term ability model, updating their comprehensive ability radar chart across dimensions such as strength, stability, rotation, reaction, and footwork, achieving guidance that combines "macro" and "micro" perspectives. Users can not only solve immediate problems but also gain a clear, data-driven understanding of their overall technical level and progress trends.

[0056] In this embodiment of the invention, the method further includes: obtaining target guidance parameters matching the current tennis movement training based on the identified movement type and / or problem diagnosis results, and projecting a laser guidance route or area according to the target guidance parameters to guide the trainee to perform the next round of training for the current tennis movement. Obtaining the target guidance parameters matching the current tennis movement training includes: acquiring the trainee's key physiological data; calibrating the laser guidance parameters of the guidance device based on the key physiological data; and extracting the target guidance parameters matching the trainee's current tennis movement training from the laser guidance parameters.

[0057] This invention proactively provides assistance based on continuous AI diagnosis. According to the identified movement type and / or problem diagnosis results, when the trainee repeatedly exhibits the same training problem, it projects a laser guidance route or area to guide the trainee in the next round of training for the current tennis movement. For example, when the AI ​​detects that a user's repeated forehand shots are due to improper footwork resulting in a too-close hitting point, it automatically activates laser projection to draw an ideal adjustment range on the ground. Furthermore, when the system starts and the trainee stands in the designated position and performs a specific movement, the vision and environment acquisition unit automatically measures the trainee's height, wingspan, and other key physiological data, and creates or loads their personal profile, automatically calibrating the laser guidance parameters of the guidance device based on the individual's key physiological data. This invention achieves truly personalized guidance, automatically calibrating subsequent laser guidance parameters (such as the height of the "virtual hitting window") based on the user's physiological data, ensuring the scientific validity and effectiveness of the guidance goals. This invention elevates the role from "passive diagnosis" to "active guidance," making the AI ​​coach's role more intelligent and human-like.

[0058] For the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should understand that the embodiments of the present invention are not limited to the described order of actions, because according to the embodiments of the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to the embodiments of the present invention.

[0059] The intelligent training method and system for analyzing all tennis techniques provided in this invention can automatically identify the type of tennis movements in real-time training. It uses a biomechanical analysis model matching the movement type to perform biomechanical analysis on the current tennis movement to determine key movement parameters. These parameters are then compared with standard movement parameters corresponding to the current tennis movement in a preset expert movement database. The comparison results and key movement parameters are combined to diagnose problems in the trainee's current tennis movement training, and the diagnostic results and improvement suggestions are provided in real-time. This represents a leap from isolated functionalities to a unified platform, providing unprecedented holistic and coherent training, ensuring more comprehensive and accurate training analysis, and achieving efficient skill improvement. This invention can be used not only for structured multi-ball practice but also for near-real-game two-person rallies or high-intensity multi-ball training, analyzing every non-preset hitting motion in real time. This greatly expands the application scenarios of intelligent training equipment, upgrading it from a "practice tool" to a "competitive partner."

[0060] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are intended to be within the scope of the invention and form different embodiments. For example, any of the claimed embodiments can be used in any combination.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent training method for realizing full technical movement analysis in tennis, characterized in that, The method includes: Acquire real-time training video streams of trainees practicing tennis moves in the training venue; Extract motion features from video image frames of the training video stream, and identify the motion type of the current tennis move based on the motion features; Based on the action type, a target biomechanical analysis model matching the action type is obtained from a preset model library; The target biomechanical analysis model is invoked to perform biomechanical analysis on the current tennis motion to determine the key motion parameters of the current tennis motion. The key motion parameters are compared with the standard motion parameters corresponding to the current tennis motion in the preset expert motion database. Based on the comparison results and the key motion parameters, the trainee's current tennis motion training is diagnosed and improvement suggestions are generated. The diagnostic results and improvement suggestions for a single tennis move will be provided in real time.

2. The method according to claim 1, characterized in that, After providing real-time feedback on the diagnostic results and improvement suggestions for a single tennis move, the method further includes: Key motion parameters, problem diagnosis results, and improvement suggestions for a single tennis move are stored in the current trainee's personal ability profile database. Based on the key motion parameters, problem diagnosis results, and improvement suggestions for each tennis move collected from the individual ability profile database, a comprehensive ability radar chart and key motion parameter change curves are generated.

3. The method according to claim 1, characterized in that, The method further includes: Based on the identified movement type and / or problem diagnosis results, target guidance parameters matching the current tennis movement training are obtained, and laser guidance routes or areas are projected according to the target guidance parameters to guide the trainee to follow the guidance for the next round of training of the current tennis movement.

4. The method according to claim 3, characterized in that, The acquisition of target guidance parameters that match the current tennis movement training includes: Obtain key physiological data from trainees; The laser guidance parameters of the guidance device are calibrated based on the key physiological data, and target guidance parameters that match the trainee's current tennis movement training are extracted from the laser guidance parameters.

5. The method according to claim 1, characterized in that, When the tennis action type is a serve, the step of calling the target biomechanical analysis model to perform biomechanical analysis on the current tennis action to determine the key action parameters of the current tennis action includes: The key joints of the human body in the video image frames corresponding to the serving action are identified to extract three-dimensional skeleton points, and a human motion dynamic model is constructed based on the temporal changes of each three-dimensional skeleton point. Based on the extraction results of each three-dimensional skeletal point, the hip-shoulder separation angle formed by hip rotation and shoulder rotation during the movement cycle is calculated, and the trunk twisting amplitude and the body's power transmission efficiency are determined according to the hip-shoulder separation angle. Calculate the angle changes and angular velocity timing of each three-dimensional skeletal point during the movement process. Analyze the order of force exertion and force transmission efficiency of each joint from the lower limbs and trunk to the upper limbs based on the angle changes and angular velocity timing. Determine whether the order of force exertion of the limbs is correct based on the order of force exertion and force transmission efficiency of each joint. Identify the tennis ball in the video image frame corresponding to the serve action, fit the three-dimensional motion trajectory of the ball based on the continuous motion coordinates of the tennis ball, compare the three-dimensional motion trajectory of the ball with the preset standard trajectory, and obtain the trajectory spatial offset and the spatial error of the highest point of the ball throw. The system identifies the three-dimensional spatial coordinates of the tennis ball and the hitting tool at the moment of impact, locates the actual hitting position, and compares the actual hitting position with the standard ideal hitting point to calculate the deviation of the hitting point position.

6. The method according to claim 1, characterized in that, When the tennis motion is a forehand, the step of calling the target biomechanical analysis model to perform biomechanical analysis on the current tennis motion to determine the key motion parameters of the current tennis motion includes: Identify the time frame of the single swing backwards limit in the video image frame corresponding to the forehand action. In the current time frame, with the trainee's torso center axis as the spatial reference plane, extract the horizontal offset distance and vertical downward sinking distance of the racket head relative to the spatial reference plane. Calculate the combined vector length through a spatial vector synthesis algorithm to obtain the maximum backwards amplitude. Acquire the vertical displacement time-series data and three-axis angular velocity time-series data of the racket head throughout the entire swing process. Identify the timestamp T1 of the lowest point of the swing and the timestamp T2 of the contact point in the corresponding video image frame of the forehand action. Calculate the average value of the instantaneous vertical acceleration of the racket head in the interval from T1 to T2 using differential calculation based on the vertical displacement time-series data and the three-axis angular velocity time-series data, and obtain the vertical acceleration of the racket head from the lowest point to the contact point.

7. An intelligent training system for analyzing all tennis technical movements, characterized in that, The system includes a physical structure body and a vision and environment acquisition unit, a motion execution and interaction unit, and a platform-based AI processing center deployed on the physical structure body. The vision and environment acquisition unit is used to acquire training video streams of trainees practicing tennis moves in the training field in a non-contact manner in real time and transmit them to the platform-based AI processing center. A platform-based AI processing center is used to execute the intelligent training method for realizing the analysis of all technical movements in tennis as described in any one of claims 1-6; The motion execution and interaction unit is used to visualize the problem diagnosis results and improvement suggestions of a single tennis move, which are fed back in real time by the platform-based AI processing center.

8. The intelligent training system for realizing full technical motion analysis in tennis according to claim 7, characterized in that, The vision and environment acquisition unit includes at least one 3D depth camera and at least one high-speed camera, or a combination of a 3D depth camera, a high-speed camera, and a wide-angle camera, for use as a training video stream when a trainee performs tennis action training in a training venue.

9. The intelligent training system for realizing full technical motion analysis in tennis according to claim 7, characterized in that, The system also includes: The laser projection unit is used to project laser guidance paths or areas onto the training area to guide the trainee in following the guidance for the next round of training of the current tennis movement.

10. The intelligent training system for realizing full technical motion analysis in tennis according to claim 7, characterized in that, The system also includes a dedicated ball supply unit for serve training; The ball supply unit for serve training includes: The storage and gravity ball supply module includes a ball storage compartment and a ball delivery slide, which is used to transport tennis balls in the ball storage compartment to the release buffer through the ball delivery slide using the principle of gravity; A buffer to be released, used to load and position tennis balls for free-fall training from the ball release chute; A tennis ball release device is used to control the release of a tennis ball from a buffer to be released.

Images