A sports trajectory data analysis system based on AI recognition

Abstract

The present invention discloses a sports trajectory data analysis system based on AI recognition, which relates to the technical field of data analysis. The equipment mid-bounds analysis module constructs a risk coefficient Fxxs through analysis, preliminarily determines the net-touching risk that the ball may have during the transmission process, and provides a more accurate evaluation for the later stage by optimizing the safety range. After the net-touching risk analysis, the equipment out-of-bounds analysis module constructs a trajectory influence coefficient Gyxs based on the training data set and combines it with the training analysis model to accurately fit the training level evaluation index Sgzs, which helps to further evaluate the performance of athletes in training to ensure the training effect and safety. The feedback module compares and analyzes the preset evaluation threshold U with the training level evaluation index Sgzs, comprehensively judges the training status, and provides targeted feedback based on the comparison results to help athletes understand the training effect in a timely manner and adjust training strategies and goals.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a sports trajectory data analysis system based on AI recognition. Background Art

[0002] In the field of intelligent sports training, AI-based recognition systems are becoming an important tool for optimizing sports training and improving athlete performance. This field covers multiple aspects such as the collection, analysis and feedback of sports data. In particular, intelligent systems used in table tennis training optimize training effects by accurately capturing motion trajectories and athlete status.

[0003] In current table tennis training, traditional training methods often rely on the coach's observation and experience, lacking scientific quantitative analysis. This method has some defects, such as it is difficult to accurately measure the details of the athlete's movements and the trajectory of the ball, and the recording and analysis of training data are often not comprehensive and timely. In addition, during the training process, athletes often hit the ball or it goes out of bounds. The analysis of this aspect is not in-depth enough, which can easily lead to a bottleneck period in training, making it difficult to overcome the phenomenon of net touching or going out of bounds. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a sports trajectory data analysis system based on AI recognition, which solves the problems in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a sports trajectory data analysis system based on AI recognition, including an image module, a data acquisition module, an equipment mid-bounds analysis module, an equipment out-of-bounds analysis module and a feedback module;

[0006] The image module is used to use an image acquisition device to shoot the entire table tennis training process of the athlete to obtain a video stream, and to intercept continuous image frames from the video stream, and to pre-process the continuous image frames to remove noise and perform geometric correction;

[0007] The data acquisition module is used to combine AI recognition technology to identify human objects and motion trajectories in continuous image frames, so as to record relevant equipment motion data information and athlete receiving and serving status data information, and generate a training data set;

[0008] The equipment mid-range analysis module is used to extract features from the training data set of the athlete during the training process to obtain the hitting angle Jqjd, the landing point distance Ldjz, the flight height difference Fxgc and the initial speed of the hitting Jcsz, and to construct the risk coefficient Fxxs under the corresponding hitting state by associating the hitting angle Jqjd and the landing point distance Ldjz. If the risk coefficient Fxxs does not fall within the safe range [ a ， b ] , initially judge that there is a risk of the ball touching the net during the transmission process, and optimize the safety range through continuous image frames [ a ， b ] ;

[0009] The equipment out-of-bounds analysis module is used to construct the trajectory influence coefficient Gyxs according to the training data set after feature extraction after analyzing whether there is a risk of net contact, and to fit and obtain the training level evaluation index Sgzs in combination with the trained training analysis model;

[0010] The feedback module is used to pre-set an evaluation threshold U, and compare and analyze the training level evaluation index Sgzs with the evaluation threshold U to comprehensively judge the status of this table tennis training.

[0011] Preferably, the image module includes an image recording unit and a processing unit;

[0012] The image recording unit is used to use an image acquisition instrument to capture the entire table tennis training process of the athlete from multiple angles, and to extract continuous image frames from the acquired video stream to ensure that the entire training is fully recorded, and to sequentially sort the continuous image frames in a corresponding order according to the data information in the image and the shooting order, so as to detect the motion trajectory of the human object and the motion trajectory of the equipment in the continuous image frames;

[0013] The processing unit is used to utilize the wavelet denoising technology to remove the noise component in the wavelet domain, and to use histogram equalization to adjust the brightness, stretch the image range, and adjust the image contrast; and at the same time compress the image data in PNG format.

[0014] Preferably, the training data set includes relevant equipment motion data information and athlete serve and receive status data information;

[0015] The data acquisition module includes a first acquisition unit and a second acquisition unit;

[0016] The first acquisition unit is used to acquire relevant equipment motion data information of the athlete during table tennis training from continuous image frames, wherein the relevant equipment motion data information includes the landing point distance Ldjz, the hitting speed Jqzs and the flying height difference Fxgc under each hitting state;

[0017] The second acquisition unit is used to acquire the athlete's serve and receive status data information during the table tennis training from continuous image frames, wherein the athlete's serve and receive status data information includes the athlete's initial hitting speed Jcsz and hitting angle Jqjd in each hitting state.

[0018] Preferably, the device boundary analysis module includes a touch analysis unit, a judgment unit and an optimization unit;

[0019] The contact analysis unit is used to obtain the risk coefficient Fxxs under the corresponding hitting state by performing linear normalization processing on the hitting angle Jqjd and the landing point distance Ldjz under the corresponding hitting state in the training data set, which is specifically obtained in the following way:

[0020] ;

[0021] In the formula, It is expressed as the risk coefficient of the i-th batting state, It is represented as the landing distance of the i-th hitting state, It is represented as the hitting angle of the i-th hitting state, It is represented as the batting speed of the i-th batting state, and are all weight values, and C represents the first correction constant.

[0022] Preferably, the judgment unit is used to combine the training data set generated in the historical stage to obtain several groups of risk coefficients Fxxs in the historical stage, and calculate the historical average risk coefficient by using the statistical mean algorithm to obtain the several groups of risk coefficients Fxxs in the historical stage. and the standard deviation of the historical risk factor ;

[0023] Based on historical average risk factor and the standard deviation of the historical risk factor , generating a safety range [ a ， b ] , the specific generated content is as follows:

[0024] ; ;

[0025] In the formula, k is a constant, and the specific value is set by the user;

[0026] By comparing the risk factor Fxxs under the corresponding hitting state with the safety range [ a ， b ] A size comparison is performed to determine whether there is a risk of the ball touching the net during transmission. The specific determination contents are as follows:

[0027] If the risk factor Fxxs under the corresponding hitting state does not fall within the safe range [ a ， b ] , it was initially determined that there was a risk of the ball touching the net during the transfer process;

[0028] If the risk factor Fxxs under the corresponding hitting state falls within the safe range [ a ， b ] , it was preliminarily judged that there was no risk of the ball touching the net during the transmission process.

[0029] Preferably, the optimization unit is used to continue to observe whether the ball is transmitted to the opponent's table according to the judgment content in the judgment unit and in combination with the continuous image frames in the corresponding hitting state, and according to the observation result, the safety range is adjusted. [ a ， b ] Optimize, the specific optimization contents are as follows:

[0030] If the ball is observed to be transferred to the opponent's table, the risk factor Fxxs under the corresponding hitting state is re-included in the safe range [ a ， b ] Inside;

[0031] If the ball is not delivered to the opponent's table, the risk factor Fxxs under the corresponding hitting state will not be included in the safety range. [ a ， b ] Inside.

[0032] Preferably, the device out-of-bounds analysis module includes a non-landing point analysis unit and a comprehensive analysis unit;

[0033] The non-landing point analysis unit is used to perform linear normalization processing on the flight height difference Fxgc and the initial speed Jcsz of the ball in the training data set, and map the corresponding data values ​​in the interval [ 0 ， 1 ] The trajectory influence coefficient Gyxs is calculated and then obtained according to the following formula:

[0034] ;

[0035] In the formula, i=1, 2, 3, ..., n, n represents the number of hits of a single player, It is expressed as the flight height difference of the i-th hitting state, Expressed as the average flight altitude difference, It is represented as the initial speed of the ball in the i-th batting state, is the average initial speed of the ball, and All are weight values.

[0036] Preferably, the comprehensive analysis unit is used to input the risk factor Fxxs and the trajectory influence coefficient Gyxs into the trained training analysis model, and after linear normalization processing, fit to obtain the training level evaluation index Sgzs:

[0037] ;

[0038] In the formula, Represented as the average risk coefficient in the current training process, and are all weight values, and Q represents the second correction constant.

[0039] Preferably, the feedback module includes a comparison and analysis unit and a state feedback unit;

[0040] The comparison and analysis unit is used to pre-set the evaluation threshold U, and compare and analyze it with the training level evaluation index Sgzs to comprehensively judge the current table tennis training status. The specific contents are as follows:

[0041] If the training level evaluation index Sgzs is greater than the evaluation threshold U, it means that the athlete's training performance has not reached the qualified target. At this time, according to the analysis results, the training method is adjusted to improve the technical and physical training, and the coach provides specific feedback to the athlete to correct the technical movements and provide support and guidance in training;

[0042] If the training level evaluation index Sgzs is equal to the evaluation threshold U, it means that the athlete's training performance has reached the qualified goal. At this time, the current training plan will continue to be implemented, the training intensity and content will be maintained, and the training effect will be evaluated in detail to check whether fine-tuning is needed;

[0043] If the training level evaluation index Sgzs is less than the evaluation threshold U, it means that the athlete's training performance has reached the qualified target. At this time, short-term goals will be set for the athlete to gradually increase the training challenge.

[0044] Preferably, the state feedback unit is used to provide real-time feedback information to coaches and athletes based on the comparison results in the comparison and analysis unit to help understand the current training status, while recording training data and feedback information, and conducting long-term analysis and optimization of training performance.

[0045] The present invention provides a sports trajectory data analysis system based on AI recognition, which has the following beneficial effects:

[0046] (1) The image module captures the entire table tennis training process of the athletes to ensure that every detail is captured. The data acquisition module combines AI recognition technology to accurately identify the movement trajectory of the athletes and the ball in the image and records detailed movement data, such as the movement status of the equipment and the status of receiving and serving. The generated training data set provides a solid data foundation for subsequent analysis. The equipment mid-bounds analysis module constructs the risk coefficient Fxxs through analysis to preliminarily determine the net contact risk that the ball may have during the transmission process, and optimizes the safety range to provide a more accurate evaluation for the later stage. After the net contact risk analysis, the equipment out-of-bounds analysis module constructs the trajectory influence coefficient Gyxs based on the training data set and combines it with the training analysis model to accurately fit the training level evaluation index Sgzs, which helps to further evaluate the performance of the athletes in training to ensure the training effect and safety. The feedback module compares and analyzes the preset evaluation threshold U with the training level evaluation index Sgzs, comprehensively judges the training status, and provides targeted feedback based on the comparison results to help athletes understand the training effect in a timely manner and adjust training strategies and goals. By analyzing the training status, the system can optimize the training plan in a targeted manner, improve the athletes' techniques and strategies, and improve the scientificity and effectiveness of the training. In short, the system is committed to identifying and preventing net contact and out-of-bounds problems, providing athletes and coaches with more scientific data support and optimization suggestions, thereby improving training results and competitive levels.

[0047] (2) By comparing the risk factor Fxxs under the corresponding hitting state with the generated safety range [ a ， b ] By comparing, the system can preliminarily determine whether there is a risk of the ball touching the net during the transmission process, and at the same time, combined with continuous image frames, further observe whether the ball is transmitted to the opponent's table. By optimizing the safety range according to the observation results, the risk assessment standard can be adjusted in real time to make the risk assessment more accurate. This optimization mechanism can dynamically adapt to changes in the actual training environment and improve the accuracy of risk prediction during training. Through re-evaluation, the system can provide coaches and athletes with scientific training adjustment suggestions so that athletes can correctly review in future training. For example, when the risk factor Fxxs is re-included in the safety range, it can be considered that the current training state has reached a good safety level. On the contrary, the training strategy needs to be adjusted to reduce the risk of net contact. Through real-time risk assessment and optimization feedback, coaches can adjust training methods and strategies in a timely manner, and athletes can also improve technical movements. This data-driven feedback mechanism helps to improve the quality and effect of training, reduce risks in training, and improve the overall competitive level. In summary, the system provides a scientific and dynamic way to monitor and manage net contact risks in table tennis training by combining historical data to generate a safety range and optimizing risk assessment in real time. The analysis of historical data ensures the reliability of risk assessment, while real-time observation and optimization enhance the system's adaptability to actual training situations.

[0048] (3) The training level evaluation index Sgzs is a comprehensive indicator that takes into account the risk factor Fxxs and the trajectory influence coefficient Gyxs. ​​It can reflect the athlete's performance in training in many aspects. The comparison and analysis unit provides targeted training feedback by comparing and analyzing the training level evaluation index Sgzs with the preset evaluation threshold U. Based on the comparison results, the system can identify whether the athlete's performance has reached the expected goal and provide specific training suggestions. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a block diagram of a sports trajectory data analysis system based on AI recognition in the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] Example 1

[0052] See also Figure 1, the present invention provides a sports trajectory data analysis system based on AI recognition, including an image module, a data acquisition module, an equipment mid-bounds analysis module, an equipment out-of-bounds analysis module and a feedback module;

[0053] The image module is used to use an image acquisition device to shoot the entire table tennis training process of the athlete to obtain a video stream, and to intercept continuous image frames from the video stream, and to pre-process the continuous image frames to remove noise and perform geometric correction;

[0054] The data acquisition module is used to combine AI recognition technology to identify human objects and motion trajectories in continuous image frames, so as to record relevant equipment motion data information and athlete receiving and serving status data information, and generate a training data set;

[0055] The equipment mid-range analysis module is used to extract features from the training data set of the athlete during the training process to obtain the hitting angle Jqjd, the landing point distance Ldjz, the flight height difference Fxgc and the initial speed of the hitting Jcsz, and to construct the risk coefficient Fxxs under the corresponding hitting state by associating the hitting angle Jqjd and the landing point distance Ldjz. If the risk coefficient Fxxs does not fall within the safe range [ a ， b ] , initially judge that there is a risk of the ball touching the net during the transmission process, and optimize the safety range through continuous image frames [ a ， b ] ;

[0056] The equipment out-of-bounds analysis module is used to construct the trajectory influence coefficient Gyxs according to the training data set after feature extraction after analyzing whether there is a risk of net contact, and to fit and obtain the training level evaluation index Sgzs in combination with the trained training analysis model;

[0057] The feedback module is used to pre-set an evaluation threshold U, and compare and analyze the training level evaluation index Sgzs with the evaluation threshold U to comprehensively judge the status of this table tennis training.

[0058] During the operation of this system, the image module uses high-quality image acquisition equipment to shoot the entire table tennis training process of the athletes, obtain continuous video streams, and capture continuous image frames from them. By preprocessing these image frames, the system can provide accurate motion image data, laying a solid foundation for subsequent analysis. This comprehensive monitoring allows every training detail to be recorded and analyzed so that coaches and athletes can watch training replays. The data acquisition module, combined with AI recognition technology, can further accurately identify the human objects and movement trajectories in the image frames, and record related equipment movement data and the athlete's receiving and serving status data. This not only generates a detailed training data set, but also can track the athlete's performance in real time, providing reliable data support for training adjustment and optimization. The equipment mid-boundary analysis module obtains key training data through feature extraction. Based on this data, the system constructs the risk factor Fxxs under the hitting state and can detect the risk of net contact. By optimizing continuous image frames, the system can dynamically adjust the safety range, further improving the prediction accuracy of net contact events, thereby reducing potential risks in training. After judging the risk of net contact, the equipment out-of-bounds analysis module constructs the trajectory influence coefficient Gyxs and combines it with the training analysis model to fit the training level evaluation index Sgzs. This quantitative evaluation method can objectively reflect the athlete's training level and provide clear feedback information for coaches and athletes to help them adjust their training strategies according to their actual performance. The feedback module sets an evaluation threshold and compares and analyzes the training level evaluation index Sgzs with the threshold. The system can comprehensively judge the training status and make corresponding adjustment suggestions. This feedback mechanism can not only correct deficiencies in training in a timely manner, but also promote the continuous improvement and enhancement of training effects. In short, the system not only improves the scientificity and accuracy of training, but also provides athletes with quantitative data support and optimization suggestions, thereby effectively improving training effects and competitive levels.

[0059] Example 2

[0060] Please refer to Figure 1 ,Specifically: the image module includes an image recording unit and a processing unit;

[0061] The image recording unit is used to use an image acquisition instrument to capture the entire table tennis training process of the athlete from multiple angles, and to extract continuous image frames from the acquired video stream to ensure that the entire training is fully recorded, and to sequentially sort the continuous image frames in a corresponding order according to the data information in the image and the shooting order, so as to detect the motion trajectory of the human object and the motion trajectory of the equipment in the continuous image frames;

[0062] The processing unit is used to utilize wavelet denoising technology to remove noise components in the wavelet domain, and to use histogram equalization to adjust brightness, stretch the image range, and adjust the image contrast; at the same time, the image data is compressed in PNG format, which is beneficial for compressing the image data without losing any information.

[0063] The training data set includes relevant equipment movement data information and athlete receiving and serving status data information;

[0064] The data acquisition module includes a first acquisition unit and a second acquisition unit;

[0065] The first acquisition unit is used to acquire relevant equipment motion data information of the athlete during table tennis training from continuous image frames, wherein the relevant equipment motion data information includes the landing point distance Ldjz, the hitting speed Jqzs and the flying height difference Fxgc under each hitting state;

[0066] The second acquisition unit is used to acquire the athlete's serve and receive status data information during the table tennis training from continuous image frames, wherein the athlete's serve and receive status data information includes the athlete's initial hitting speed Jcsz and hitting angle Jqjd in each hitting state.

[0067] Linear normalization of the training data set can standardize the relevant parameters within it and eliminate the impact of different measurement units or magnitudes on data analysis. This processing method enables the data to be compared and analyzed on a unified scale, further improving the accuracy of risk assessment.

[0068] In this embodiment, the image recording unit of the image module ensures that the whole process of the athlete's table tennis training is fully recorded by shooting from multiple angles. The interception and sorting of continuous image frames in the video stream ensure that every detail in the training process is accurately recorded, which helps to accurately analyze the athlete's movements and the use of the equipment. The processing unit uses wavelet denoising technology to effectively remove image noise to ensure the clarity of image data, and the histogram equalization technology adjusts the image brightness and contrast, making the image easier to analyze. At the same time, the compression technology of the PNG format maintains the integrity of the image data without losing information, and provides high-quality image data for subsequent analysis. The first acquisition unit obtains the relevant equipment movement data information in the athlete's training from the continuous image frames. This information helps to deeply understand the performance of the equipment in training and provides accurate data support for performance evaluation in the batting state. The second acquisition unit collects the athlete's receiving and serving state data information. These data help evaluate the athlete's batting technique and training effect, and reveal the athlete's performance and progress in different batting states. The training data set includes relevant equipment movement data information and athlete receiving and serving state data information. By comprehensively analyzing these data, the influence of various factors in the training process on the ball's motion trajectory and the athlete's performance can be further fully understood.

[0069] Example 3

[0070] Please refer to Figure 1 ,Specifically: the device boundary analysis module includes a touch analysis unit, a judgment unit and an optimization unit;

[0071] The contact analysis unit is used to obtain the risk coefficient Fxxs under the corresponding hitting state by performing linear normalization processing on the hitting angle Jqjd and the landing point distance Ldjz under the corresponding hitting state in the training data set, which is specifically obtained in the following way:

[0072] ;

[0073] In the formula, It is expressed as the risk coefficient of the i-th batting state, It is represented as the landing distance of the i-th hitting state, It is represented as the hitting angle of the i-th hitting state, It is represented as the batting speed of the i-th batting state, and are weight values, where 0 < ≤1,0< ≤1, and + =1, C is represented as the first correction constant.

[0074] The above-mentioned landing point distance Ldjz refers to the distance between the landing point of the table tennis ball on the table tennis table and the net, which can be monitored and obtained by a laser rangefinder;

[0075] The hitting angle Jqjd can be monitored and obtained through an angle sensor;

[0076] The ball speed Jqzs refers to the speed of the ball during the transmission process, which can be monitored and obtained by a laser speed meter;

[0077] In table tennis training, the net in the middle of the table is an important part of the playing field. It divides the table into two areas and helps divide the playing areas of the two players. The players must receive and hit the ball in their respective half of the court.

[0078] In this embodiment, the touch analysis unit helps to accurately assess the net touch risk under different hitting states by calculating the risk coefficient Fxxs under the corresponding hitting state, providing a scientific basis for risk control in the training process. The weight value and correction constant used in the formula can be adjusted according to the actual situation, so that the risk assessment is more in line with the actual training environment and the performance of the players. The system can detect and warn of possible net touch risks in real time, and convey risk information to athletes and coaches through a feedback mechanism. In this way, coaches can adjust training strategies in time, and athletes can also improve their movements in future training processes to reduce the risk of net touch, thereby improving training effects and safety. By calculating and analyzing the risk coefficient Fxxs, the system can provide detailed feedback to help coaches and athletes identify potential problems in training. According to the risk assessment results, the training plan can be adjusted in a targeted manner, optimize technical movements, and reduce the occurrence of net touch events. The system provides scientific support for the net touch risk assessment during table tennis training through accurate risk coefficient calculation and dynamic optimization mechanism.

[0079] Example 4

[0080] Please refer to Figure 1 Specifically: the judgment unit is used to combine the training data set generated in the historical stage to obtain several groups of risk coefficients Fxxs in the historical stage, and calculate the historical average risk coefficient by using the statistical mean algorithm to obtain the several groups of risk coefficients Fxxs in the historical stage. and the standard deviation of the historical risk factor ;

[0081] Based on historical average risk factor and the standard deviation of the historical risk factor , generating a safety range [ a ， b ] , the specific generated content is as follows:

[0082] ; ;

[0083] In the formula, k is a constant, usually with a value of 1-3 to correspond to different confidence levels, and the specific value is set by the user;

[0084] By comparing the risk factor Fxxs under the corresponding hitting state with the safety range [ a ， b ] A size comparison is performed to determine whether there is a risk of the ball touching the net during transmission. The specific determination contents are as follows:

[0085] If the risk factor Fxxs under the corresponding hitting state does not fall within the safe range [ a ， b ] , it was initially determined that there was a risk of the ball touching the net during the transfer process;

[0086] If the risk factor Fxxs under the corresponding hitting state falls within the safe range [ a ， b ] , it was preliminarily judged that there was no risk of the ball touching the net during the transmission process.

[0087] Among them, net contact is used to describe the event that the ball touches the net during flight.

[0088] The optimization unit is used to continue to observe whether the ball is transmitted to the opponent's table according to the judgment content in the judgment unit and in combination with the continuous image frames in the corresponding hitting state, and according to the observation result, the safety range is adjusted. [ a ， b ] Optimize, the specific optimization contents are as follows:

[0089] If the ball is observed to be transferred to the opponent's table, the risk factor Fxxs under the corresponding hitting state is re-included in the safe range [ a ， b ] Inside;

[0090] If the ball is not delivered to the opponent's table, the risk factor Fxxs under the corresponding hitting state will not be included in the safety range. [ a ， b ] Inside.

[0091] In this embodiment, the judgment unit obtains several groups of risk coefficients Fxxs in the historical stage by combining the training data set generated in the historical stage, and performs statistical analysis on these risk coefficients Fxxs, including calculating their mean and standard deviation, to provide a scientific basis for generating a safety range. This historical data analysis method can effectively capture long-term risk trends and provide a reference for real-time risk assessment. Based on the historical average risk coefficient and standard deviation, the generated safety range can be dynamically adjusted, and the specific confidence level is set by the constant k. This method allows the user to set different confidence levels according to actual needs, thereby providing an adaptable risk assessment in different training stages or scenarios. By comparing the risk coefficient Fxxs in the corresponding hitting state with the safety range, the system can judge in real time whether there is a risk of the ball touching the net during the transmission process. If the risk coefficient Fxxs does not fall into the safety range, it is initially judged that there is a risk of touching the net; if the risk coefficient Fxxs falls into the safety range, it is judged that there is no risk of touching the net. This real-time judgment mechanism helps coaches and athletes to instantly understand the risk situation in training. The optimization unit optimizes the safety range based on the judgment results of the judgment unit and the observation data of continuous image frames. By observing whether the ball is successfully transmitted to the opponent's table, the system can adjust the safety range to make it more accurate and adapt to the actual training situation. If the ball is transmitted to the opponent's table, the risk factor Fxxs is re-included in the safety range; if the ball is not transmitted to the opponent's table, the risk factor Fxxs is excluded from the safety range. This optimization process ensures the real-time and accuracy of risk assessment. The system can provide scientific safety range and risk assessment results based on historical data and real-time observations, which enables coaches and athletes to make targeted adjustments based on specific risk data, optimize training movements and strategies, and improve the safety and effectiveness of training. The system's dynamic safety range generation and optimization mechanism can provide data-based decision support for coaches and athletes, reduce training interference caused by net contact or other risks, and thus achieve more efficient and safer training.

[0092] Example 5

[0093] Please refer to Figure 1 ,Specifically: the device out-of-bounds analysis module includes a non-landing point analysis unit and a comprehensive analysis unit;

[0094] The non-landing point analysis unit is used to perform linear normalization processing on the flight height difference Fxgc and the initial speed Jcsz of the ball in the training data set, and map the corresponding data values ​​in the interval [ 0 ， 1 ] The trajectory influence coefficient Gyxs is calculated and then obtained according to the following formula:

[0095] ;

[0096] In the formula, i=1, 2, 3, ..., n, n represents the number of hits of a single player, It is expressed as the flight height difference of the i-th hitting state, Expressed as the average flight altitude difference, It is represented as the initial speed of the ball in the i-th batting state, is the average initial speed of the ball, and are weight values, where 0 < ≤1,0< ≤1, and + =1.

[0097] The above-mentioned flight height difference Fxgc refers to the difference between the highest position and the lowest position of the ball during the transmission process under the corresponding hitting condition, which can be monitored and obtained by a laser rangefinder;

[0098] The initial speed Jcsz of the ball can be monitored and obtained by a radar gun or a laser speed meter.

[0099] In this embodiment, the non-landing point analysis unit can map the data values ​​of these parameters to a unified interval by linearly normalizing the flight height difference Fxgc and the initial speed of the ball Jcsz in the training data set. This processing method eliminates the influence of different units and magnitudes on data analysis and improves the accuracy and consistency of the trajectory influence coefficient calculation. By using the normalized flight height difference Fxgc and the initial speed of the ball Jcsz, the trajectory influence coefficient Gyxs is calculated so that the system can accurately identify and judge whether the ball is likely to go out of bounds during the transmission process, thereby achieving early warning of possible out-of-bounds problems. By analyzing the trajectory influence coefficient Gyxs of each batting state, the system can provide coaches and athletes with detailed information about the risk of going out of bounds, so that the technical movements of the players can be improved in a targeted manner, the occurrence of going out of bounds can be reduced, and the quality of training can be improved.

[0100] Among them, the ball fails to land on the opponent's table tennis table, that is, the ball flies directly out of the opponent's playing area after flying over the net. This situation is usually called out-of-bounds.

[0101] Example 6

[0102] Please refer to Figure 1 Specifically: the comprehensive analysis unit is used to input the risk coefficient Fxxs and the trajectory influence coefficient Gyxs into the trained training analysis model, and after linear normalization processing, fit to obtain the training level evaluation index Sgzs:

[0103] ;

[0104] In the formula, Represented as the average risk coefficient in the current training process, and are all weight values, Q represents the second correction constant, where 0 < ≤1,0< ≤1, and + =1.

[0105] Among them, the average risk coefficient in the current training process Refers to the average risk factor during the entire current training process;

[0106] It should be noted that a convolutional neural network is used to construct a basic model, and the basic model is trained and tested with relevant equipment motion data information and athlete receiving and serving state data information, and the trained basic model is used as a state recognition model, and feature information in the state recognition model is obtained respectively, and the state recognition model is trained and tested with the obtained feature information, and the trained state recognition model is used as a training analysis model;

[0107] The feedback module includes a comparison and analysis unit and a state feedback unit;

[0108] The comparison and analysis unit is used to pre-set the evaluation threshold U, and compare and analyze it with the training level evaluation index Sgzs to comprehensively judge the current table tennis training status. The specific contents are as follows:

[0109] If the training level evaluation index Sgzs is greater than the evaluation threshold U, it means that the athlete's training performance has not reached the qualified target. At this time, according to the analysis results, the training method is adjusted to improve the technical and physical training, and the coach provides specific feedback to the athlete to correct the technical movements and provide support and guidance in training;

[0110] If the training level evaluation index Sgzs is equal to the evaluation threshold U, it means that the athlete's training performance has reached the qualified target. At this time, the current training plan will continue to be implemented, the training intensity and content will be maintained to ensure that the athlete can continue to reach the target level, and the training effect will be evaluated in detail to check whether fine-tuning is needed to further optimize the training effect;

[0111] If the training level evaluation index Sgzs is less than the evaluation threshold U, it means that the athlete's training performance has reached the qualified target. At this time, short-term goals will be set for the athlete to gradually increase the training challenges and further improve the athlete's skills and abilities.

[0112] The state feedback unit is used to provide real-time feedback information to coaches and athletes based on the comparison results in the comparison and analysis unit to help understand the current training status, while recording training data and feedback information to facilitate subsequent tracking of the athlete's progress and long-term analysis and optimization of training performance.

[0113] In the present embodiment, the comparison and analysis unit compares and analyzes the training state of the athlete according to the evaluation threshold value U and the training level evaluation index Sgzs, and can effectively judge the training state of the athlete. And according to the training level evaluation index Sgzs numerical value, the system will provide specific training improvement measures. For example, if the training level evaluation index Sgzs is greater than the threshold, the system will suggest adjusting the training method to improve the technology and physical fitness, and ensure that the performance of the athlete reaches the expected goal; the state feedback unit provides real-time feedback information to the coach and the athlete according to the result in the comparison and analysis unit, and this real-time feedback helps the coach and the athlete to quickly understand the current training state and adjust the training plan and method in time. At the same time, the system records the training data and feedback information, and provides a basis for the follow-up tracking of the progress of the athlete and long-term analysis. This long-term tracking and analysis can help further optimize the training effect and ensure that the athlete continues to improve in technology and physical fitness. By recording and analyzing the training data, the system can identify the shortcomings in the training, provide scientific data support, and help the coach and the athlete make targeted improvements in future training. This data-driven training optimization method can further improve the effect and efficiency of training, so that the athlete can obtain more significant progress in a shorter period of time. The comprehensive analysis unit and feedback module provide a comprehensive assessment of the athlete's training performance and optimization suggestions through precise calculation and real-time feedback of the training level evaluation index Sgzs. This system can not only provide targeted improvement measures based on the specific training status, but also realize real-time feedback and long-term tracking, helping coaches and athletes make scientific, data-driven decisions in training, thereby further improving the training effect and the athletes' competitive level.

[0114] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A sports trajectory data analysis system based on AI recognition, characterized by: It includes an image module, a data acquisition module, an instrument mid-bounds analysis module, an instrument out-of-bounds analysis module and a feedback module; The image module is used to use an image acquisition device to shoot the entire table tennis training process of the athlete to obtain a video stream, and to intercept continuous image frames from the video stream, and to pre-process the continuous image frames to remove noise and perform geometric correction; The data acquisition module is used to combine AI recognition technology to identify human objects and motion trajectories in continuous image frames, so as to record relevant equipment motion data information and athlete receiving and serving status data information, and generate a training data set; The relevant equipment motion data information includes the landing point distance Ldjz, the hitting speed Jqzs and the flying height difference Fxgc in each hitting state; the athlete receiving and serving state data information includes the athlete's initial hitting speed Jcsz and the hitting angle Jqjd in each hitting state; The equipment mid-range analysis module is used to extract features from the training data set of the athlete during the training process to obtain the hitting angle Jqjd, the landing point distance Ldjz, the flight height difference Fxgc and the initial speed of the hitting Jcsz, and to construct the risk coefficient Fxxs under the corresponding hitting state by associating the hitting angle Jqjd and the landing point distance Ldjz. If the risk coefficient Fxxs does not fall within the safe range , initially judge that there is a risk of the ball touching the net during the transmission process, and optimize the safety range through continuous image frames ; The device boundary analysis module includes a touch analysis unit, a judgment unit and an optimization unit; The contact analysis unit is used to obtain the risk coefficient Fxxs under the corresponding hitting state by performing linear normalization processing on the hitting angle Jqjd and the landing point distance Ldjz under the corresponding hitting state in the training data set, which is specifically obtained in the following way: ； In the formula, It is expressed as the risk coefficient of the i-th batting state, It is represented as the landing distance of the i-th hitting state, It is represented as the hitting angle of the i-th hitting state, It is represented as the batting speed of the i-th batting state, and They are all weight values, and C represents the first correction constant; The judgment unit is used to obtain several groups of risk coefficients Fxxs in the historical stage by combining the training data set generated in the historical stage, and to obtain the historical average risk coefficient by respectively calculating the several groups of risk coefficients Fxxs in the historical stage through the statistical averaging algorithm. and the standard deviation of the historical risk factor ; Based on historical average risk factor and the standard deviation of the historical risk factor , generating a safety range , the specific generated content is as follows: ； ； In the formula, k is a constant, and the specific value is set by the user; By comparing the risk factor Fxxs under the corresponding hitting state with the safety range A size comparison is performed to determine whether there is a risk of the ball touching the net during transmission. The specific determination contents are as follows: If the risk factor Fxxs under the corresponding hitting state does not fall within the safe range , it was initially determined that there was a risk of the ball touching the net during the transfer process; If the risk factor Fxxs under the corresponding hitting state falls within the safe range , it was initially determined that there was no risk of the ball touching the net during the transfer process; The optimization unit is used to continue to observe whether the ball is transmitted to the opponent's table according to the judgment content in the judgment unit and in combination with the continuous image frames in the corresponding hitting state, and according to the observation result, the safety range is adjusted. Optimize, the specific optimization contents are as follows: If the ball is observed to be transferred to the opponent's table, the risk factor Fxxs under the corresponding hitting state is re-included in the safe range Inside; If the ball is not delivered to the opponent's table, the risk factor Fxxs under the corresponding hitting state will not be included in the safety range. Inside; The equipment out-of-bounds analysis module is used to construct the trajectory influence coefficient Gyxs according to the training data set after feature extraction after analyzing whether there is a risk of net contact, and to fit and obtain the training level evaluation index Sgzs in combination with the trained training analysis model; The device out-of-bounds analysis module includes a non-landing point analysis unit and a comprehensive analysis unit; The non-landing point analysis unit is used to perform linear normalization processing on the flight height difference Fxgc and the initial speed Jcsz of the ball in the training data set, and map the corresponding data values ​​in the interval The trajectory influence coefficient Gyxs is calculated and obtained by the following formula: ； In the formula, i=1, 2, 3, ..., n, n represents the number of hits of a single player, It is expressed as the flight height difference of the i-th hitting state, Expressed as the average flight altitude difference, It is represented as the initial speed of the ball in the i-th batting state, is the average initial speed of the ball, and All are weight values; The comprehensive analysis unit is used to input the risk factor Fxxs and the trajectory influence coefficient Gyxs into the trained training analysis model, and after linear normalization processing, fit to obtain the training level evaluation index Sgzs: ； In the formula, Represented as the average risk coefficient in the current training process, and All are weight values, Q represents the second correction constant; The feedback module is used to pre-set an evaluation threshold U, and compare and analyze the training level evaluation index Sgzs with the evaluation threshold U to comprehensively judge the status of this table tennis training.

2. The sports trajectory data analysis system based on AI recognition according to claim 1, characterized in that: The image module includes an image recording unit and a processing unit; The image recording unit is used to use an image acquisition instrument to capture the entire table tennis training process of the athlete from multiple angles, and to extract continuous image frames from the acquired video stream to ensure that the entire training is fully recorded, and to sequentially sort the continuous image frames in a corresponding order according to the data information in the image and the shooting order, so as to detect the motion trajectory of the human object and the motion trajectory of the equipment in the continuous image frames; The processing unit is used to utilize the wavelet denoising technology to remove the noise component in the wavelet domain, and to use histogram equalization to adjust the brightness, stretch the image range, and adjust the image contrast; and at the same time compress the image data in PNG format.

3. The sports trajectory data analysis system based on AI recognition according to claim 1, characterized in that: The training data set includes relevant equipment movement data information and athlete receiving and serving status data information; The data acquisition module includes a first acquisition unit and a second acquisition unit; The first acquisition unit is used to acquire relevant equipment motion data information of the athlete during table tennis training from continuous image frames; The second acquisition unit is used to acquire data information on the player's serving and receiving status during table tennis training from continuous image frames.

4. The sports trajectory data analysis system based on AI recognition according to claim 1, characterized in that: The feedback module includes a comparison and analysis unit and a state feedback unit; The comparison and analysis unit is used to pre-set the evaluation threshold U, and compare and analyze it with the training level evaluation index Sgzs to comprehensively judge the current table tennis training status. The specific contents are as follows: If the training level evaluation index Sgzs is greater than the evaluation threshold U, it means that the athlete's training performance has not reached the qualified target. At this time, according to the analysis results, the training method is adjusted to improve the technical and physical training, and the coach provides specific feedback to the athlete to correct the technical movements and provide support and guidance in training; If the training level evaluation index Sgzs is equal to the evaluation threshold U, it means that the athlete's training performance has reached the qualified goal. At this time, the current training plan will continue to be implemented, the training intensity and content will be maintained, and the training effect will be evaluated in detail to check whether fine-tuning is needed; If the training level evaluation index Sgzs is less than the evaluation threshold U, it means that the athlete's training performance has reached the qualified target. At this time, short-term goals will be set for the athlete to gradually increase the training challenge.

5. The sports trajectory data analysis system based on AI recognition according to claim 4, characterized in that: The state feedback unit is used to provide real-time feedback information to coaches and athletes based on the comparison results in the comparison and analysis unit to help understand the current training status, while recording training data and feedback information, and conducting long-term analysis and optimization of training performance.

Images