A sports visual data review and analysis system and method based on ball hitting trigger

Abstract

This invention discloses a system and method for reviewing and analyzing motion visual data based on ball-hitting triggers, relating to the field of motion multimodal data synchronization and intelligent analysis technology. The system includes a racket trigger module, a head-mounted visual acquisition module, and a data processing terminal, which work together to form an inseparable core architecture of "ball-hitting trigger - data acquisition - synchronized review and analysis." Specifically, the head-mounted visual acquisition module acquires the athlete's first-person perspective scene in real time, annotates it with a high-precision timestamp, and streams it to the data processing terminal in real time, without setting up a large local cache to reduce device load; the racket trigger module acquires ball-hitting related inertial navigation data, completing ball-hitting detection and timestamp generation and uploading through two independent and non-conflicting methods; the data processing terminal caches the received data, triggers deduplication and time-series alignment, and the time-series alignment adopts the nearest original frame matching strategy to preserve the authenticity of the original data and avoid data distortion. This invention enables high-precision synchronization of hitting action and visual data, supports both automated analysis by the host computer and manual review / autonomous analysis by the user, and features rapid data transfer and stable storage capabilities. It solves the technical problems of multi-source data timing misalignment, chaotic trigger signals, high load on head-mounted devices, and poor adaptability in existing technologies. It can be flexibly adapted to various racket sports such as table tennis, badminton, and tennis, providing data support for professional training in racket sports, while reserving space for parameter adjustment in subsequent research and development.

Description

Technical Field

[0001] This invention relates to the fields of motion sensing, multimodal data synchronous acquisition and data processing, and in particular to a system and method for accurately reviewing and flexibly analyzing motion visual data by relying on hardware-based ball-hitting timestamp marking, real-time transmission from the head-mounted device, and long-term cache storage on a host computer. Background Technology

[0002] In training for racket-based sports such as table tennis, badminton, and tennis, the moment of impact is the core time reference for analyzing athletes' visual behavior. Only by accurately aligning the moment of impact can we truly reflect the visual prediction before impact, the stability of the line of sight at the moment of impact, and the visual feedback process after impact, providing reliable data support for training optimization and movement correction.

[0003] Most existing visual acquisition and analysis systems divide data into time periods using continuous acquisition, fixed time segments, or post-event manual video annotation. This fails to use the actual hitting action as a unified time anchor, easily leading to temporal misalignment between eye-tracking data, scene image data, and the hitting action. Consequently, visual behavior analysis results deviate from the actual motion state, significantly reducing accuracy. Furthermore, existing systems generally lack anti-shake and deduplication mechanisms for the hitting trigger signal. Vibrations and slight impacts during racket impact can easily generate duplicate markings or false trigger signals, further compromising the reliability of the data temporal sequence.

[0004] In terms of data architecture design, traditional head-mounted devices with large-capacity local caches significantly increase device power consumption and overall weight, affecting athlete comfort and making them unsuitable for prolonged training. Furthermore, the lack of a long buffering mechanism in the host computer makes it easy to lose crucial pre-hit prediction data, failing to fully reconstruct the visual state throughout the entire hitting process. Currently, the industry lacks a reliable technical solution that uses the hitting time point as the core, with real-time data acquisition and uploading from the head-mounted device and centralized caching processing on the host computer, while also possessing trigger deduplication and high-precision timing alignment functions. This makes it difficult to meet the high-precision, automated visual analysis needs of professional sports training. Summary of the Invention

[0005] To address the problems of existing technologies failing to use the moment of impact as a precise time reference, resulting in misalignment between visual data and the timing of the impact action, frequent false triggering of trigger signals, unreasonable data caching architecture, and easy loss of critical data, this invention provides a motion visual data synchronization analysis system and method based on impact triggering. This system achieves high-precision timing synchronization between multimodal motion visual data and the impact action, improving analysis accuracy and system practicality.

[0006] The system includes a racket triggering module, a head-mounted vision acquisition module, and a data processing terminal.

[0007] The racket trigger module is used to detect the hitting action and output the corresponding hitting time signal in real time. It does not participate in data acquisition and storage, thus reducing the module load and power consumption.

[0008] The head-mounted visual acquisition module is used to acquire scene image data in real time, providing raw data for multi-dimensional visual analysis. At the same time, it wirelessly streams various types of data to the data processing terminal in real time, without setting up a large local cache, thus achieving device lightweighting.

[0009] The data processing terminal includes a global ring-shaped long buffer unit, which is used to continuously cache multimodal data streams for a preset duration to ensure that the complete historical data before the ball is hit is not overwritten or lost. At the same time, it performs preliminary regularization and verification on the cached data stream, providing a complete and reliable data foundation for subsequent time alignment and segment extraction, and realizing the coordinated connection between data storage and preliminary processing.

[0010] Furthermore, the data processing terminal also includes a trigger deduplication unit, which is used to perform anti-shake and deduplication processing on the received ball-hitting time signal, filter out false triggers and duplicate trigger signals, obtain a unique and valid ball-hitting timestamp, and ensure the accuracy of the timing reference.

[0011] Furthermore, the data processing terminal also includes a timing alignment and segment extraction unit, which is configured to accurately extract data segments from a global circular long buffer for a preset period before and after the shot, using a unique and valid shot timestamp as the core timing reference, thereby achieving precise timing alignment between visual data and the shot time point.

[0012] Furthermore, the timing alignment unit performs time offset compensation based on the local timestamp carried by the head-mounted data and the ball-hitting trigger timestamp, eliminating the deviation caused by hardware transmission and clock differences, and realizing unified timing calibration of multi-source data.

[0013] Furthermore, the system performs clock drift correction through periodic wireless clock synchronization packets, ensuring that the wireless synchronization accuracy between the racket triggering module and the head-mounted vision acquisition module is no less than ±1ms, meeting the requirements for high-precision synchronization analysis.

[0014] Furthermore, due to the differences in sampling frequencies among the eye-tracking acquisition unit, scene acquisition unit, and posture detection unit, the output times of each data frame cannot precisely coincide with the moment of impact. The timing alignment unit adopts the nearest original frame matching strategy, directly selecting the real acquired image with the timestamp closest to the moment of impact to complete the alignment with the posture data. No image interpolation is performed, no virtual images are generated, and the original acquired data is retained throughout the process, eliminating data distortion and ensuring the authenticity of timing matching and the reliability of analysis.

[0015] Furthermore, this invention provides two implementation schemes for ball-hitting action detection and timestamp generation. The two schemes can be implemented independently and do not conflict with each other, as detailed below: Option 1 (Local Detection Scheme): The first microcontroller of the racket trigger module presets a ball impact threshold to analyze the impact and vibration signals collected by the inertial detection unit. After identifying a valid ball-hitting action, it immediately generates a high-precision ball-hitting timestamp. Only this timestamp and the device identifier are transmitted in real time to the data processing terminal via the first wireless communication unit. The trigger deduplication unit of the data processing terminal completes the subsequent deduplication processing. This scheme can reduce the wireless transmission pressure and the computational load of the data processing terminal, and improve the real-time performance and timing accuracy of the ball-hitting timestamp generation.

[0016] Option 2 (Host Computer Detection Scheme): The first microcontroller of the racket trigger module is only responsible for controlling the inertial detection unit to collect raw impact and vibration data, and uploading the raw data to the data processing terminal in real time through the first wireless communication unit. The data processing terminal also includes a hit detection unit, which receives the raw data uploaded by the inertial detection unit. The user can select the corresponding sports type through the data processing terminal, and the system automatically matches the hit impact threshold corresponding to the sports type to complete the identification of valid hit actions and generate a unique and valid hit timestamp. Then, the trigger deduplication unit completes the subsequent deduplication processing. This scheme can reduce the performance requirements of the first microcontroller of the racket trigger module, reduce the size and power consumption of the racket trigger module, and improve the system's adaptability to different racket holding sports, allowing switching of sports types without modifying the racket hardware.

[0017] Accordingly, this invention also proposes a method for synchronous analysis of motion visual data triggered by ball impact, specifically including the following steps: The head-mounted vision acquisition module collects scene image data in real time and wirelessly streams various types of data marked with local timestamps to the data processing terminal. The data processing terminal continuously caches and stores the received multimodal data through a global circular long buffer unit, preserving complete data before and after the shot; The inertial detection unit of the racket trigger module collects the impact and vibration signals of the ball in real time, and uses one of the following two methods to complete the detection of the ball hitting action and the uploading of the ball hitting time signal: Option 1: The first microcontroller of the racket trigger module presets the impact threshold of the ball, and immediately generates a ball-hitting timestamp after recognizing a valid ball-hitting action, and uploads the timestamp to the data processing terminal; Option 2: The first microcontroller of the racket trigger module uploads the raw impact and vibration data collected by the inertial detection unit to the data processing terminal. The data processing terminal selects the motion type, matches the corresponding impact threshold, completes the identification of the effective hitting action, and generates a hitting timestamp. By triggering the deduplication unit, the hit time signal is deduplicated and anti-shake processed to eliminate false and duplicate signals and determine the unique valid hit timestamp. Using the valid shot timestamp as the timing reference, combined with timestamp offset compensation and wireless clock synchronization correction, and the nearest original frame matching strategy, multi-source data alignment is completed. Data segments corresponding to the time period are extracted from the global circular long buffer to achieve high-precision timing alignment between visual data and the moment of shot. Visual behavior analysis was performed on the standardized data fragments after synchronization to extract quantitative training indicators.

[0018] Furthermore, this invention uses the ball-hitting event as a unified time anchor point, combining real-time transmission from the head-mounted display, centralized long buffering on the host computer, trigger deduplication, and high-precision wireless timing alignment technology. Through two flexibly selectable ball-hitting detection schemes, while ensuring the system is lightweight, low-power, and comfortable to wear, it relies on real raw data to complete timing matching, realizing automated and highly reliable motion vision synchronous analysis centered on the ball-hitting time point. The corresponding ball-hitting detection scheme can be selected according to the actual application scenario (such as timing accuracy requirements, hardware cost, and multi-motion adaptation requirements).

[0019] Furthermore, the present invention also has the ability to quickly jump to and store data. It can stably store the data segments before and after the ball is hit. Users can manually call up the stored data segments and quickly jump to any target time point without relying on the automatic analysis function of the host computer. Data review and autonomous analysis can be completed by manual operation alone, which is suitable for diverse use scenarios such as simple review and manual analysis.

[0020] This invention uses the actual time of ball impact as a unified temporal benchmark, completely resolving the problem of misalignment between eye-tracking data, scene data, posture data, and the timing of ball impact actions, thus fundamentally improving the accuracy of motion vision analysis. Simultaneously, without the need for manual annotation and segmentation of video data, the system can automatically complete the extraction of ball impact segments and simultaneous analysis of multi-source data, significantly reducing training data processing time and improving training analysis efficiency. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall system architecture of the present invention. Reference numerals: 100-Racket trigger module, 101-First microcontroller, 102-Inertial detection unit, 103-First wireless communication unit, 104-First power module; 200-Head-mounted vision acquisition module, 201-Second microcontroller, 202-Camera unit, 203-Second wireless communication unit, 204-Second power module; 300-Data processing terminal, 301-Wireless receiving unit, 302-Buffer unit, 303-Data processing unit, 304-Analysis unit, 305-Synchronization unit, 306-Ball hit detection unit.

[0022] Figure 2 This is a flowchart illustrating the data flow and timing synchronization process of this invention. Detailed Implementation

[0023] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0024] This invention discloses a motion visual data synchronization analysis system and method based on ball-hitting trigger. The system consists of three main parts: a racket trigger module 100, a head-mounted visual acquisition module 200, and a data processing terminal 300. Each module works together to complete the entire process of data acquisition, transmission, caching, synchronization, and analysis.

[0025] The racket trigger module 100 is installed inside the racket handle or frame, fitting the racket structure and not affecting normal hitting. The module has a built-in first microcontroller 101, inertial detection unit 102, first wireless communication unit 103 and first power module 104. The components work together to complete the acquisition, processing, transmission and time synchronization of the hitting signal. The first power module 104 provides stable power to all components within the module, ensuring continuous and stable operation. The inertial detection unit 102 uses a three-axis accelerometer paired with a gyroscope to collect impact and vibration signals from the ball in real time, accurately sensing the ball-hitting action. The first microcontroller 101, as the core control unit of the racket triggering module 100, is responsible for coordinating the data acquisition rhythm and preliminary signal processing of the inertial detection unit 102 (executing corresponding logic according to the selected ball-hitting detection scheme), as well as controlling the first wireless communication unit 103 to complete signal uploading. It also has the ability to receive periodic wireless synchronization packets sent by the data processing terminal 300 through the first wireless communication unit 103 and complete its own clock calibration, ensuring the timing synchronization accuracy between the racket triggering module 100, the head-mounted vision acquisition module 200, and the data processing terminal 300. The first wireless communication unit 103 uses a low-power wireless transmission protocol to transmit the signals output by the first microcontroller 101 (ball-hitting timestamp or raw impact data) to the data processing terminal 300 in real time, while forwarding time synchronization-related instructions and data to ensure the real-time performance and stability of data transmission and time synchronization. The racket trigger module 100 does not preset specific hit types and can adapt to the hit detection needs of various racket sports such as table tennis, badminton, and tennis.

[0026] The head-mounted visual acquisition module 200 is worn on the athlete's head, featuring an ergonomic design for lightweight and comfortable wear. The module integrates a second microcontroller 201, a camera unit 202, a second wireless communication unit 203, and a second power module 204. The camera unit 202 integrates scene acquisition functionality, capturing real-time first-person perspective training scene footage from the athlete. All acquired data is annotated with a local high-precision timestamp and streamed in real-time to the data processing terminal 300 via the second wireless communication unit 203. The head-mounted device does not have a large-capacity local cache; it is only responsible for continuous acquisition and real-time transmission, minimizing device load, weight, and power consumption. It also forwards time synchronization-related instructions and data.

[0027] The data processing terminal 300 integrates a wireless receiving unit 301, a buffer unit 302, a data processing unit 303, an analysis unit 304, and a synchronization unit 305. In Scheme Two, the data processing terminal 300 additionally includes a ball-hitting detection unit 306. The wireless receiving unit 301 continuously receives real-time data streams from the head-mounted device and signals (timestamps or raw data) transmitted from the racket, ensuring continuous data reception. The buffer unit 302 employs a global ring-shaped long buffer structure, continuously caching the most recent multimodal complete data stream. New data automatically overwrites older data, permanently retaining the latest real-time data, ensuring that key data in the pre-hitting prediction stage is not lost, providing complete data support for subsequent analysis.

[0028] In addition, the head-mounted vision acquisition module 200 has reserved expansion interfaces to be compatible with new acquisition components, adapt to the diverse data acquisition needs in different training scenarios, and adding new acquisition components does not change the core working logic of this module or its collaboration mechanism with other modules.

[0029] The data processing unit 303 integrates trigger deduplication and timing management functions, and executes dedicated trigger optimization logic: it presets a minimum effective trigger interval of 100ms to 200ms, and retains only the first effective hit timestamp within the same time window; at the same time, it sets graded impact intensity thresholds to filter out false trigger signals generated by slight collisions and racket face self-vibration; for fast and continuous legal hit actions, it independently retains the corresponding timestamps to avoid the loss of effective training data, and finally outputs a clean and unique hit timestamp sequence.

[0030] Synchronization unit 305 first calibrates the clocks of the racket, headset, and terminal hardware via periodic wireless synchronization packets, stabilizing the wireless timing synchronization accuracy within ±1ms. Then, it performs fine-grained offset compensation based on the original timestamps of each acquisition unit, unifying the timing reference across multiple data sources. To address frame time deviations caused by differences in sampling rates among different acquisition units, it employs the nearest original frame matching method, selecting the most accurate timestamp from the actual eye movement, scene, and posture data at the point of impact for alignment. This process avoids generating virtual images and performing interpolation, ensuring the original authenticity of the data.

[0031] Analysis unit 304 uses the effective shot timestamp as an anchor point to extract data segments from buffer unit 302 for preset time periods before and after the shot, automatically segmenting and organizing the data. Analysis unit 304 incorporates an object detection and recognition model to perform real-time target recognition on the scene captured by head-mounted vision, locating key targets such as the ball, court boundaries, opponents, and equipment. It combines eye-tracking data to determine the correspondence between the athlete's gaze target and scene objects. Based on this, it calculates quantitative indicators such as the athlete's gaze landing point before and after the shot, ball-gazing duration, line-of-sight deviation, head posture coordination, and ball position matching degree at the moment of shot, generating a visualized training analysis report and evaluation data to provide a basis for coach guidance and athlete self-optimization.

[0032] In Scheme 1 (Local Detection Scheme), both the ball-hitting action detection and timestamp generation are completed locally in the racket triggering module 100. The specific process is as follows: The first microcontroller 101 of the racket triggering module 100 presets the ball-hitting impact threshold (which can be preset according to a single motion type or multiple motion type thresholds for quick switching later). The inertial detection unit 102 collects the impact and vibration signals of the racket in real time. The first microcontroller 101 compares the collected signals with the preset thresholds and filters out valid ball-hitting actions (excluding invalid signals such as slight vibrations and non-ball-hitting collisions).

[0033] After the first microcontroller 101 recognizes a valid hitting action, it immediately generates a high-precision hitting timestamp. Only this timestamp and the device identifier of the racket trigger module are sent to the data processing terminal 300 in real time through the first wireless communication unit 103. No original impact data is transmitted, thus reducing the pressure on wireless transmission and the power consumption of the module.

[0034] After the wireless receiving unit 301 of the data processing terminal 300 receives the ball-hitting timestamp, the data processing unit 303 performs anti-shake and deduplication processing to generate a unique and valid ball-hitting timestamp. The subsequent timing synchronization, segment extraction and data analysis processes are consistent with the core process of the system.

[0035] The advantages of this solution are: extremely low latency in generating the ball-hitting timestamp, high timing accuracy, small amount of wireless data transmission, low requirements for the computing power of the data processing terminal 300, strong anti-interference ability, and suitability for scenarios with high timing accuracy requirements and long-term training of a single sport type.

[0036] In this second scheme (host computer detection scheme), the racket trigger module 100 only collects raw impact data. The ball-hitting action detection and timestamp generation are completed by the data processing terminal 300. The specific process is as follows: The first microcontroller 101 of the racket trigger module 100 does not execute the ball-hitting action recognition logic. It only controls the inertial detection unit 102 to collect raw impact and vibration data in real time. After marking the raw data with a local timestamp, it is uploaded to the data processing terminal 300 in real time through the first wireless communication unit 103. Since there is no need to execute complex recognition logic, the first microcontroller 101 can be a model with lower performance, smaller size and lower power consumption, further reducing the size and power consumption of the racket trigger module 100.

[0037] After receiving the raw impact data, the wireless receiving unit 301 of the data processing terminal 300 transmits it to the newly added ball-hitting detection unit 306. The user can select the corresponding sport type (such as table tennis, badminton, tennis) through the interactive interface of the data processing terminal 300. The ball-hitting detection unit 306 automatically matches the ball-hitting impact threshold corresponding to the sport type, analyzes the raw impact data, identifies valid ball-hitting actions, and generates a high-precision ball-hitting timestamp.

[0038] The generated shot timestamp is transmitted to the data processing unit 303. After anti-shake and deduplication processing is completed, it enters the subsequent timing synchronization, segment extraction and data analysis process, which is completely consistent with the subsequent process of Scheme 1.

[0039] The advantages of this solution are: the racket trigger module 100 has low hardware requirements, small size, and low power consumption. It does not require modification of the racket hardware and can adapt to various racket-holding sports simply by switching the sports type through the host computer. In the future, the detection algorithm and threshold can be iteratively optimized through the host computer, resulting in low maintenance costs. It is suitable for training in multiple sports types and scenarios with strict requirements on the size of racket hardware.

[0040] The overall workflow of this invention is as follows, and the corresponding ball-hitting detection scheme can be selected according to requirements: S1. After the device is powered on and initialized, the head-mounted vision acquisition module continuously acquires first-view image data, uploads it in real time after marking the timestamp, and the data processing terminal receives the data and stores it in the global ring long buffer unit in real time. S2. Select the ball-hitting detection scheme to complete the ball-hitting action detection and signal upload: Option 1: The racket trigger module identifies valid shots through the first microcontroller, generates and sends a shot timestamp to the data processing terminal; Option 2: The racket trigger module uploads the raw impact data, and the data processing terminal matches the motion type, identifies valid hits, and generates a hit timestamp through the hit detection unit; S3, the data processing terminal eliminates false and duplicate trigger signals through trigger deduplication and timing management, and at the same time completes multi-hardware clock calibration and timing offset compensation; S4. Employ the nearest original frame matching strategy to achieve accurate alignment of multi-source data and automatically extract standard data segments before and after the ball is hit; S5. Perform visual behavior analysis on the synchronized and normalized data segments, and output quantitative indicators and visual training evaluation results.

[0041] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention; any modifications, equivalent substitutions, optimizations and improvements made within the spirit and principles of the present invention should be covered within the protection scope of the present invention.

Claims

1. A system and method for reviewing and analyzing motion visual data triggered by ball impact, characterized in that, It includes a racket trigger module, a head-mounted vision acquisition module, and a data processing terminal; The racket triggering module and the head-mounted vision acquisition module are both wirelessly connected to the data processing terminal, and are used to transmit ball-related signals and multimodal visual data to the data processing terminal. The racket triggering module is used to collect data and timestamps related to hitting the ball and upload them to the data processing terminal. The head-mounted vision acquisition module is used to acquire multimodal vision data and stream it to the data processing terminal in real time, without setting up a local large-capacity cache. The data processing terminal is used to cache, deduplicatize, align timing, and analyze the received signals and data to achieve precise synchronous analysis of multimodal visual data and hitting actions.

2. The system according to claim 1, characterized in that, The racket triggering module has a built-in power module, an inertial detection unit, a first microcontroller, and a first wireless communication unit. The power module supplies power to the components within the racket trigger module; The inertial detection unit is used to collect impact and vibration signals from the ball in real time (i.e., inertial navigation data), which is used to detect ball-hitting events. The first microcontroller is responsible for coordinating the data acquisition rhythm of the inertial detection unit and the signal uploading logic of the first wireless communication unit; The first wireless communication unit is used to upload ball-related signals to the data processing terminal.

3. The system according to claim 1, characterized in that, The head-mounted visual acquisition module collects multimodal visual data, and all collected data is labeled with a local high-precision timestamp. The head-mounted visual acquisition module wirelessly streams the collected data to the data processing terminal in real time.

4. The system according to claim 1, characterized in that, The racket triggering module, the head-mounted vision acquisition module, and the data processing terminal use periodic wireless clock synchronization packets to correct clock drift, ensuring that the wireless synchronization accuracy between the racket triggering module and the head-mounted vision acquisition module is no less than ±1ms.

5. The system according to claim 1, characterized in that, The data processing terminal (300) also has the functions of fast data jump, storage and manual analysis. It can stably store the data segments before and after the ball is hit after the time sequence is aligned. Users can manually call up the stored data segments and quickly jump to any target time point. It does not rely on the upper computer automatic analysis function of the data processing terminal (300). Data review and autonomous analysis can be completed by manual operation, which is suitable for the use scenario of simple data review and manual analysis.

6. A method for synchronous analysis of motion visual data triggered by ball impact, characterized in that, Includes the following steps: S1. Real-time acquisition of scene image data is achieved through a head-mounted vision acquisition module. All acquired data is labeled with a local high-precision timestamp and the acquired data is wirelessly streamed to the data processing terminal in real time. S2. The inertial detection unit of the racket trigger module collects the impact and vibration signals of the ball in real time, and uses one of two schemes to complete the detection of the ball hitting action and the generation of the ball hitting timestamp, and uploads the ball hitting related signals to the data processing terminal through the first wireless communication unit; S3, the data processing terminal eliminates false and duplicate trigger signals through trigger deduplication and timing management, and at the same time completes multi-hardware clock calibration and timing offset compensation; S4. Employ the nearest original frame matching strategy to achieve accurate alignment of multi-source data and automatically extract standard data segments before and after the ball is hit; S5. Perform visual behavior analysis on the synchronized and normalized data segments, and output quantitative indicators and visual training evaluation results.

7. The system according to claim 1, characterized in that, The first method (Scheme 1) for the racket trigger module to complete the detection of the hitting action and the generation of the hitting timestamp is as follows: The first microcontroller of the racket trigger module presets the hitting impact threshold, analyzes the impact and vibration signals (inertial navigation data) collected by the inertial detection unit, and immediately generates a high-precision hitting timestamp after identifying a valid hitting action. Only this timestamp and the device identifier are uploaded to the data processing terminal through the first wireless communication unit, and the trigger deduplication unit of the data processing terminal completes the subsequent deduplication processing.

8. The system according to claim 1, characterized in that, The second method (Scheme 2) for the racket trigger module to complete the detection of the hitting action and the generation of the hitting timestamp is as follows: The data processing terminal further includes a hitting detection unit. The first microcontroller of the racket trigger module does not execute the hitting action recognition logic, but only controls the inertial detection unit to collect the original impact and vibration data (inertial navigation data), and uploads the original data to the data processing terminal through the first wireless communication unit after annotating it with a local timestamp. The hitting detection unit receives the original data, and the user can select the corresponding sports type through the data processing terminal. The hitting detection unit automatically matches the hitting impact threshold corresponding to the sports type, identifies the valid hitting action and generates a high-precision hitting timestamp, and then transmits it to the trigger deduplication unit for deduplication processing.

9. The method according to claim 6, characterized in that, In step S2, the two schemes for detecting the hitting action and generating the hitting timestamp can be implemented independently and do not conflict with each other; Scheme 1 is that the racket trigger module completes the hitting recognition and timestamp generation locally, and Scheme 2 is that the data processing terminal completes the hitting recognition and timestamp generation.

10. The method according to claim 6, characterized in that, The analysis unit has a built-in object detection and recognition model; using the effective shot timestamp obtained in step S4 as the anchor point, it extracts data segments from the cached multimodal data stream before and after the shot within a preset time window; it performs object detection and recognition only on the scene image data within the segment, identifying one or more targets among the ball, field, opponent, and sports equipment, and completes the motion visual analysis.

Images