A basketball tactic demonstration method and system

Abstract

An embodiment of the present invention provides a basketball tactic demonstration method and system, belonging to the field of intelligent sports technology. The method includes: responding to a tactic demonstration request signal of a user, identifying a corresponding tactic serial number and player information for executing the tactic; calling a on-court movement model of the corresponding player based on the player information for executing the tactic as a first model; calling a defensive model of the corresponding team based on the current on-court opponent information as a second model; performing tactic process model training under the corresponding tactic serial number based on the first model and the second model to obtain a corresponding tactic simulation result; performing visualization processing on the tactic simulation result, and pushing the visualization result to the user terminal.

Description

Technical Field

[0001] The present invention relates to the field of intelligent sports technology, and particularly to a basketball tactic demonstration method and a basketball tactic demonstration system. Background Art

[0002] In the process of basketball games and training, the formulation and execution of tactics have a decisive impact on the game results. Traditional tactic demonstration methods mainly rely on the coach's experience, draw the player's running positions and passing lines on a tactic board, and analyze them in combination with game videos. However, it is difficult to quantify the success rate of tactic execution in this way, nor can it accurately simulate the performance of players in different tactic environments. With the progress of data analysis technology, some basketball tactic analysis systems have begun to use 2D or 3D animations for tactic demonstration, attempting to improve the intuitiveness of tactic understanding, but these systems still have many technical limitations. For example, most existing tactic demonstrations are preset fixed paths and cannot be adjusted individually in combination with the actual movement data of players. At the same time, these systems usually only simulate the running positions and passes of the attacking side, without considering the dynamic changes of the defensive side, resulting in a lack of real confrontation environment in tactic analysis and unable to accurately evaluate the success rate of tactic execution.

[0003] In actual games, the effects of the same tactic executed by different players often vary significantly. For example, in a non-defended state, some players have a faster movement speed, but may be greatly affected under high-pressure defense, resulting in a decrease in the efficiency of tactic execution. In addition, the shooting hit rate of players in different positions is also closely related to their historical performance and defensive pressure. Existing tactic demonstration systems cannot provide personalized tactic success rate predictions, resulting in a lack of accurate data support for coaches when formulating tactics. Similarly, factors such as the defensive positions, stealing, and blocking abilities of the defensive side have an important impact on the result of tactic execution, but existing tactic simulation methods often lack a defensive model and can only present fixed tactic demonstrations, unable to dynamically simulate the interaction between the attacking and defensive sides in the game. Due to the lack of accurate modeling of individual player characteristics and defensive environments, existing systems are difficult to provide scientific tactic optimization solutions, making tactic demonstration and analysis still rely on subjective experience judgment.

[0004] In summary, the main problems of existing basketball tactic demonstration methods are the lack of tactic success rate prediction based on personalized data and the failure to dynamically simulate the impact of defense on tactic execution. To address these problems, a new basketball tactic demonstration solution needs to be proposed. Summary of the Invention

[0005] The purpose of the embodiments of the present invention is to provide a basketball tactic demonstration method and system to at least solve the problems of the existing basketball tactic demonstration method, namely, the lack of tactic success rate prediction based on personalized data and the failure to dynamically simulate the impact of defense on tactic execution.

[0006] To achieve the above object, a first aspect of the present invention provides a basketball tactic demonstration method, the method comprising: responding to a tactic demonstration request signal of a user, identifying a corresponding tactic serial number and player information for executing the tactic; invoking a on-court movement model of the corresponding player based on the player information for executing the tactic as a first model; invoking a defensive model of the corresponding team based on the opponent information on the current court as a second model; performing training on a tactic process model under the corresponding tactic serial number based on the first model and the second model to obtain a corresponding tactic simulation result; performing visualization processing on the tactic simulation result and pushing the visualization result to the user terminal.

[0007] Optionally, the on-court movement model is obtained by training based on training samples constituted by on-court movement historical data of the corresponding player; wherein, the on-court movement historical data includes: the movement speed of the corresponding player in a non-defended state, the movement speed of bypassing a pick-and-roll, the actual on-court positions at each moment, and the shooting hit rate data at each point under different defensive pressure levels.

[0008] Optionally, the training rules of the on-court movement model include: performing data preprocessing on the on-court movement historical data of the target player; based on the on-court movement historical data after data preprocessing, taking the movement speed of the corresponding player in a non-defended state and the movement speed of bypassing a pick-and-roll as first training samples; taking the shooting hit rate data at each point of the corresponding player under different defensive pressure levels as second training samples; performing model training based on the Hidden Markov Model algorithm and the first training samples to obtain an on-court position prediction model of the corresponding player; performing model training based on the Logistic Regression algorithm and the second training samples to obtain a shooting hit rate prediction model of the corresponding player; performing coupling of the on-court position prediction model and the shooting hit rate prediction model to obtain the on-court movement model of the corresponding player.

[0009] Optionally, the performing model training based on the Hidden Markov Model algorithm and the first training samples to obtain an on-court position prediction model of the corresponding player includes: taking the tactic states under different tactic serial numbers as hidden variables of the Hidden Markov Model algorithm, taking the actual on-court positions of the corresponding player at each moment as observation data of the Hidden Markov Model algorithm to obtain corresponding first training samples; setting a state transition matrix, an observation probability matrix and an initial state distribution of the Hidden Markov Model based on the first training samples; in the iterative process, calculating the expected value of each round of iteration based on the Forward-Backward algorithm and adjusting the state transition matrix, the observation probability matrix and the initial state distribution by maximizing the likelihood function until the convergence condition is satisfied to obtain an optimal state transition matrix and observation probability matrix; constructing a corresponding Hidden Markov Model based on the optimal state transition matrix and observation probability matrix as the on-court position prediction model of the corresponding player.

[0010] Optionally, training the model based on the logistic regression algorithm and the second training sample to obtain a shooting percentage prediction model for the corresponding player, including: performing annotations on shooting position, defensive pressure level, player shooting type, and game background information for each piece of data in the second training sample to obtain the annotated second training sample; wherein, the defensive pressure level is determined based on the distance between the defensive player and the current player and the maximum defensive height of the defensive player; the game background information is the total playing time of the current player in this game; using the annotated second training sample as the input features of the logistic regression model to perform model training; during the model training process, calculating the loss function based on maximum likelihood estimation and adjusting the parameters through gradient descent until the convergence condition is met to obtain a shooting percentage prediction model for the corresponding player.

[0011] Optionally, the training rules for the defensive model of the corresponding team include: predicting the future position distribution of the defensive player based on the hidden Markov model to calculate the state transition probability of the defensive position based on the individual defensive ability of the defensive player, defensive tactical strategy, and offensive player behavior; calculating the defensive influence factor based on the state transition probability of the defensive position; wherein, the defensive influence factor includes the degree of interference of the defensive distance on the shooting percentage, the probability of the defensive player blocking the shot, and the degree of influence of the steal on the passing success rate; using the defensive influence factor as the influence factor in the tactical execution process to simulate the decision-making changes of the offensive player under different defensive modes to obtain an initial model; adjusting the position model and defensive influence factor of each defensive player of the corresponding team in the initial model based on historical game data and combining the game situation under the corresponding historical data to adjust the model parameters to obtain the defensive model of the corresponding team.

[0012] Optionally, the tactical simulation results include: the position change process of the tactical execution player, the position change process of the defensive player, the simulation process of the basketball transfer path, and the final predicted shooting percentage value.

[0013] Optionally, performing visualization processing on the tactical simulation results and pushing the visualization results to the user side, including: generating a dynamic tactical demonstration screen based on the position change process of the tactical execution player and the position change process of the defensive player; performing real-time annotation of tactical data on the dynamic tactical demonstration screen; wherein, the tactical data includes any one or more of the running route of the player, the passing trajectory, and the shooting percentage of the real-time position; pushing the annotated dynamic tactical demonstration screen to the user side.

[0014] In a second aspect of the present invention, a basketball tactical demonstration system is provided. The system includes: an analysis unit configured to identify a corresponding tactical serial number and player information for executing a tactic in response to a tactical demonstration request signal from a user; a first model calling unit configured to call a on-court movement model of a corresponding player based on the player information for executing the tactic as a first model; a second model calling unit configured to call a defensive model of a corresponding team based on opponent information on the current court as a second model; a simulation unit configured to perform training on a tactical process model under the corresponding tactical serial number based on the first model and the second model to obtain a corresponding tactical simulation result; and a visualization unit configured to perform visualization processing on the tactical simulation result and push the visualization result to a user terminal.

[0015] On the other hand, the present invention provides a computer-readable storage medium having instructions stored thereon that, when run on a computer, cause the computer to execute the above-described basketball tactical demonstration method.

[0016] Through the above technical solution, the solution of the present invention automatically identifies a target tactic and related player information based on a user's tactical demonstration request, and calls a corresponding on-court movement model of the player and a defensive model to construct a tactical simulation system that better conforms to the actual game environment. By combining the personalized movement data of offensive players and the dynamic changes of the defensive side, this method can perform training on the tactical process model and generate more accurate tactical simulation results. Compared with traditional tactical demonstrations that only rely on fixed paths and static tactical diagrams, the present invention can dynamically adjust player running positions, passing paths, and shooting selections, making the tactical demonstration more realistic. At the same time, the tactical simulation result is pushed to the user terminal after visualization processing, providing an intuitive tactical demonstration effect, enabling coaches and players to clearly understand the tactical execution process and possible game impacts. This method not only improves the accuracy and adaptability of the tactical demonstration, but also enables the development of tactical analysis from empirical judgment to data-driven optimization, providing more scientific support for tactical formulation and player training.

[0017] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description section. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention, and constitute a part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the drawings:

[0019] Figure 1 is a flowchart of the steps of a basketball tactical demonstration method provided by an embodiment of the present invention;

[0020] Figure 2It is the system structure diagram of the basketball tactic demonstration system provided by an embodiment of the present invention. Specific Embodiment

[0021] The following will describe in detail the specific embodiments of the present invention with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for the purpose of illustrating and explaining the present invention, and are not intended to limit the present invention.

[0022] Figure 1 It is the method flow diagram of the basketball tactic demonstration method provided by an embodiment of the present invention. As Figure 1 shown, an embodiment of the present invention provides a basketball tactic demonstration method, and the method includes:

[0023] Step S10: In response to the tactic demonstration request signal of the user, identify the corresponding tactic number and the player information for executing the tactic.

[0024] Specifically, the identification of the tactic number not only includes the tactic type, but may also involve specific execution plans. For example, when the same tactic is executed by different players, there may be significant differences in their running positions, passing strategies, and shooting methods. Therefore, accurately identifying the tactic number and player information is the basis for tactic simulation. The determination of the tactic number ensures that the subsequent called tactic model matches the user's needs, making the tactic demonstration process more accurate and realistic. When identifying the player information for executing the tactic, it is necessary to comprehensively consider the player's historical data, including characteristics such as the moving speed in the undefended state, the moving method of bypassing screens, and the shooting hit rate under different defensive pressures, so as to accurately call the tactic execution method that conforms to the player's movement characteristics. In addition, real-time game data can also be combined to ensure that the player's current state (such as physical energy consumption, game performance) matches the tactic execution requirements. Compared with the traditional static tactic demonstration, this method can intelligently match player data according to the input tactic information to achieve personalized tactic simulation.

[0025] Step S20: Call the on-court movement model of the corresponding player based on the player information for executing the tactic as the first model.

[0026] Specifically, the on-court movement model is obtained by training with training samples composed of the on-court movement historical data of the corresponding player; wherein, the on-court movement historical data includes: the moving speed of the corresponding player in the undefended state, the moving speed of bypassing screens, the actual on-court positions at each moment, and the shooting hit rate data at each point under different defensive oppression levels.

[0027] In the embodiments of the present invention, the on-court movement model is obtained by training a training sample constructed based on the on-court movement historical data of the corresponding player. The training sample covers the movement characteristics of the player under different tactical execution situations to ensure that the model can accurately simulate the movement pattern of the player in a real game environment. The on-court movement historical data contains multiple key variables, including any one or more of the movement speed of the player in a non-defended state, the movement speed of bypassing a screen, the actual on-court position at each moment, and the shooting hit rate data at each point under different defensive pressure levels. These data together constitute the basis of the player's movement behavior in different scenarios, enabling the trained model to not only reflect the natural movement pattern of the player in an open position but also reflect the dynamic changes during tactical execution.

[0028] Furthermore, the movement speed in a non-defended state reflects the player's autonomous running ability without defensive pressure, while the movement speed of bypassing a screen is a key factor in offensive tactical execution, determining whether the player can quickly break away from the defender and enter the best offensive position. The actual on-court position data at each moment is used to record the running position pattern of the player in the game, providing an accurate reference for the movement trajectory in modeling. The shooting hit rate data at each point under different defensive pressure levels further supplements the individual differences during tactical execution, enabling the model to dynamically adjust the prediction of the player's shooting behavior in combination with defensive pressure, thus being closer to the real game situation.

[0029] Preferably, the training rules of the on-court movement model include: performing data preprocessing on the on-court movement historical data of the target player; based on the on-court movement historical data after data preprocessing, taking the movement speed of the corresponding player in a non-defended state and the movement speed of bypassing a screen as the first training sample; taking the shooting hit rate data at each point of the corresponding player under different defensive pressure levels as the second training sample; performing model training based on the Hidden Markov Model algorithm and the first training sample to obtain the on-court position prediction model of the corresponding player; performing model training based on the Logistic Regression algorithm and the second training sample to obtain the shooting hit rate prediction model of the corresponding player; performing coupling of the on-court position prediction model and the shooting hit rate prediction model to obtain the on-court movement model of the corresponding player.

[0030] In the embodiments of the present invention, the training rules of the on-court movement model include a series of data processing and modeling steps to ensure that the model can accurately simulate the movement patterns and shooting performances of players in different game situations. First, the on-court movement historical data of the target player is preprocessed to eliminate abnormal data and standardize the data format, so as to improve the quality and consistency of the training samples. After the data preprocessing is completed, the movement speed of the player in the undefended state and the movement speed around the pick-and-roll are used as the first training samples, and combined with the actual on-court positions of the player at each moment, to construct the basic movement pattern of the player in the offensive tactics. At the same time, the shooting hit rate data of the player at each point under different defensive pressure levels are used as the second training samples to train the shooting decision-making model, so that it can accurately predict the shooting success rate of the player under different defensive intensities.

[0031] During the model training process, the first training samples are trained based on the Hidden Markov Model (HMM) algorithm to establish an on-court position prediction model for the player. This model can learn the movement rules of the player in different tactical situations and predict their possible positions at future time steps, thus providing dynamic running position data for tactical simulation. At the same time, the second training samples are trained based on the Logistic Regression algorithm to establish a shooting hit rate prediction model, which can comprehensively consider factors such as shooting position and defensive pressure to calculate the shooting success probability of the player in different areas. Finally, the on-court position prediction model and the shooting hit rate prediction model are coupled to form a complete on-court movement model, which can not only predict the running trajectory of the player but also provide an accurate estimation of the shooting success rate in combination with the position information.

[0032] Specifically, the model training based on the Hidden Markov Model algorithm and the first training samples to obtain the on-court position prediction model for the corresponding player includes: taking the tactical states under different tactical serial numbers as the hidden variables of the Hidden Markov Model algorithm, and taking the actual on-court positions of the corresponding player at each moment as the observation data of the Hidden Markov Model algorithm to obtain the corresponding first training samples; setting the state transition matrix, observation probability matrix, and initial state distribution of the Hidden Markov Model based on the first training samples; in the iterative process, calculating the expected value of each iteration based on the forward-backward algorithm, and adjusting the state transition matrix, the observation probability matrix, and the initial state distribution by maximizing the likelihood function until the convergence condition is met to obtain the optimal state transition matrix and observation probability matrix; constructing the corresponding Hidden Markov Model based on the optimal state transition matrix and observation probability matrix as the on-court position prediction model for the corresponding player.

[0033] In the embodiments of the present invention, the tactical states under different tactical serial numbers are used as the hidden variables of the hidden Markov model, and the actual on-field positions of the players at each moment are used as the observed data, thereby constructing the first training sample. The introduction of the tactical state ensures that the model can distinguish the movement patterns of players in different tactical situations, while the selection of the observed data provides real movement trajectory information for training, enabling the model to learn the possible position distributions of players during the execution of different tactics.

[0034] After obtaining the first training sample, the core parameters of the hidden Markov model are set based on this data, including the state transition matrix, the observation probability matrix, and the initial state distribution. The state transition matrix is used to describe the probability of a player moving from one position to another between different time steps, and is defined as:

[0035] ;

[0036] where, represents the probability that the player transfers from state to state ; is the position state of the player at time step t. The observation probability matrix is used to measure the possibility of a player appearing at a specific position under a specific tactical state, and is defined as:

[0037] ;

[0038] where, is the observed position of the player; reflects the probability of observing in state . The initial state distribution determines the standing tendency of the player at the initial stage of the tactical execution, and is defined as:

[0039] ;

[0040] where, represents the probability that the player is initially in state .

[0041] During the model training process, the forward-backward algorithm is used for expectation calculation, and the state transition matrix, the observation probability matrix, and the initial state distribution are adjusted by combining the method of maximizing the likelihood function. The forward-backward algorithm optimizes the prediction of the player's movement trend by calculating the probability distribution of the current state and combining historical data and possible future states. The application of the maximum likelihood function enables the training process to gradually converge, and finally obtains the state transition matrix and the observation probability matrix that can best describe the player's movement pattern. Iterative training continuously optimizes these parameters until the convergence condition is met, that is, the prediction ability of the model reaches a stable and accurate state.

[0042] Finally, based on the optimal state transition matrix and observation probability matrix obtained from training, a hidden Markov model is constructed and used as the on-field position prediction model for players. Given the tactical serial number and game environment, this model can predict the possible positions of players at future time steps, thereby providing dynamic movement trajectory data for tactical simulation. Compared with traditional fixed-path tactical demonstrations, the present invention enables tactical simulation to dynamically adjust according to tactical changes and player habits through the hidden Markov model, improving the accuracy and personalization level of tactical demonstrations.

[0043] Based on the solution of the present invention, by introducing a hidden Markov model for player position prediction, tactical simulation evolves from a static and preset running position trajectory into an intelligent prediction model based on historical data and tactical background. This model can adapt to the movement patterns of different players in different tactical environments, making tactical simulation closer to real game situations. In addition, through the maximum likelihood optimization algorithm, the model can adaptively adjust parameters to ensure the stability and accuracy of prediction results, providing scientific data support for tactical formulation, player training, and game analysis.

[0044] Preferably, training the model based on the logistic regression algorithm and the second training sample to obtain a shooting hit rate prediction model for the corresponding player includes: performing annotation on the shooting position, defensive pressure level, player shooting type, and game background information of each data in the second training sample to obtain the annotated second training sample; wherein, the defensive pressure level is determined based on the distance between the defensive player and the current player and the maximum defensive height of the defensive player; the game background information is the total playing time of the current player in this game; using the annotated second training sample as the input feature of the logistic regression model to perform model training; during the model training process, calculating the loss function based on the maximum likelihood estimation and adjusting the parameters through gradient descent until the convergence condition is met to obtain the shooting hit rate prediction model for the corresponding player.

[0045] In the embodiment of the present invention, data annotation is performed on the second training sample to ensure that the model can accurately capture the key factors affecting the shooting hit rate. During the data annotation process, first determine the shooting position, using the shooting point (x, y) of the player as the input variable to establish the relationship between the shooting area and the hit rate. Then, calculate the defensive pressure level, which is defined based on the distance d between the defensive player and the current shooting player and the maximum defensive height h of the defensive player, indicating the degree of interference exerted by the defensive player on the shooting player. The calculation formula for the defensive pressure level can be expressed as:

[0046] ;

[0047] Among them, d represents the Euclidean distance between the defensive player and the shooting player, h represents the maximum defensive height of the defensive player, and adding 1 to the denominator avoids the problem of division by zero in the calculation. When the defensive player is relatively close and has a relatively high height, this value is larger, indicating a stronger defensive interference, which may reduce the shooting percentage.

[0048] Furthermore, the shooting type is an important factor affecting the shooting success rate, including catch-and-shoot, dribble shooting, fade-away jump shot, etc. The success rates corresponding to each shooting method are different. Therefore, the shooting type is encoded as a categorical variable. Finally, considering the game background information, especially the total playing time of the player to measure the impact of the player's physical condition on the shooting performance. As the game time increases, the player's shooting percentage may decrease due to fatigue. Therefore, is used as one of the input features. After completing the data annotation, it is used as an input feature of the logistic regression model to establish a shooting percentage prediction model. The basic mathematical model of logistic regression is expressed as: ;

[0049] Among them, represents the probability of a successful shot given the input features X (including shooting location, defensive pressure, game time, and shooting type); are the weight parameters to be trained. During the model training process, the maximum likelihood estimation algorithm is used to calculate the loss function to ensure that the model parameters can maximize the fitting degree of the shooting success rate. The loss function of logistic regression is defined as:

[0050] ;

[0051] Among them, N is the number of training samples, indicates whether the shot of the i-th sample is successful (1 means successful, 0 means unsuccessful). The weight parameter w is updated through the gradient descent algorithm, and the loss function is iteratively optimized until the convergence condition is met, that is, the prediction error of the model is reduced to the minimum and remains stable.

[0052] Furthermore, after the training is completed, the logistic regression model can output the estimated value of the shooting success rate given the shooting location, defensive pressure level, shooting type, and game background information of the player. The technical effect of this technical solution is that by constructing a data-driven shooting percentage prediction model, the tactical simulation can not only present the player's running position and shooting situation, but also predict the shooting success rate based on data analysis, thereby improving the accuracy of tactical evaluation. Compared with the traditional tactical demonstration method, this method can dynamically adjust the shooting decision by combining individual player data, making the tactical formulation more scientific and providing more targeted tactical optimization suggestions for coaches and players.

[0053] Step S30: Based on the opponent information on the current field, the defense model of the corresponding team is called as the second model.

[0054] Specifically, the training rules of the defense model of the corresponding team include: predicting the future position distribution of the defensive players based on the hidden Markov model, and calculating the state transition probability of the defense position based on the individual defense ability of the defensive players, the defense tactical strategy and the behavior of the offensive players; calculating the defense influence factor based on the state transition probability of the defense position; wherein the defense influence factor includes the interference degree of the defense distance on the shooting accuracy, the probability of the defensive player blocking and the influence degree of the steal on the passing success rate; using the defense influence factor as the influencing factor of the tactical execution process, simulating the decision changes of the offensive players under different defense modes, and obtaining the initial model; adjusting the position model and defense influence factor of each defensive player of the corresponding team in the initial model based on the historical game data, and adjusting the model parameters in combination with the actual game situation under the corresponding historical data, so as to obtain the defense model of the corresponding team.

[0055] In the embodiment of the present invention, the future position distribution of the defensive player is predicted based on the hidden Markov model to calculate the state transition probability of the defensive position. In this process, the individual defensive ability of the defensive player, the defensive tactical strategy and the behavior of the offensive player jointly determine the movement pattern of the defensive player. The state transition probability matrix is ​​used to describe the probability of the defensive player moving from one position to another between different time steps, which can be expressed as:

[0056] ;

[0057] in, represents the position of the defensive player at the current time step, is the possible position in the next time step, Represents the current offensive player's position, The tactical status is the corresponding element in the state transition probability matrix, indicating the transition probability from state i to state j. Based on the state transition probability of the defensive position, the defensive influence factor is further calculated to quantify the degree of interference of the defensive player on the shooting and passing success rates. The defensive influence factor mainly includes three aspects: the impact of the defensive distance on the shooting accuracy, the probability of the defensive player blocking, and the impact of the steal on the passing success rate. Among them, the interference degree of the defensive distance on the shooting accuracy can be calculated by the following formula:

[0058] ;

[0059] Among them, d represents the distance between the defensive player and the offensive player, and λ is a decay coefficient set empirically. When the defensive player is close to the offensive player, the interference degree is large and the shooting percentage decreases; on the contrary, when the defensive distance is far, the interference effect gradually decreases. The block probability calculation is modeled based on the height of the defensive player, the blocking tendency, and the opponent's shooting method, while the steal probability calculation is based on the position of the defensive player, the reaction speed, and the intersection area of the passing trajectory.

[0060] Furthermore, when constructing the initial model, the calculated defensive influence factor is introduced into the tactical execution process to simulate the decision-making changes of the offensive player under different defensive modes. This model can simulate the influence of man-to-man defense, zone defense, defensive rotation, and double-team strategies on the offensive player's choices. For example, when the system detects that the offensive player is facing high defensive pressure, it may predict that he is more inclined to pass the ball rather than shoot directly. To improve the accuracy of the defensive model, the initial model needs to be optimized by combining historical game data. By analyzing the actual positions, blocks, and steals of defensive players in historical games, the state transition probability and defensive influence factor of the model are corrected to make it more in line with the actual game situation.

[0061] Based on the solution of the present invention, the position of the defensive player is predicted through the hidden Markov model, and the changes in shooting and passing success rates are calculated in combination with the defensive influence factor, so that the tactical simulation not only focuses on the tactical execution of the offensive side, but also can truly reproduce the tactical adjustments of the defensive side. Compared with the traditional static tactical demonstration, this method improves the authenticity of the tactical simulation, enabling it to dynamically calculate the tactical execution success rate of the offensive side under different defensive tactical configurations, thereby providing more accurate data support for tactical optimization.

[0062] Step S40: Based on the first model and the second model, perform the training of the tactical process model corresponding to the tactical serial number to obtain the corresponding tactical simulation result.

[0063] Specifically, the tactical simulation result includes: the position change process of the tactical execution player, the position change process of the defensive player, the simulation process of the basketball transfer path, and the predicted value of the final shooting percentage.

[0064] In the embodiments of the present invention, based on the first model and the second model, the tactical process model corresponding to the tactical serial number is trained to obtain the corresponding tactical simulation result. In this process, the first model, that is, the on-court movement model of the player, predicts the movement pattern of the player under a specific tactic by learning features such as the player's historical movement trajectory, speed in the unguarded state, the way of bypassing screens, and shooting percentage. The second model, that is, the defensive model of the team, combines the positions of the defensive players, the tendency to steal, the shot-blocking ability, and the defensive strategy to simulate the response measures of the defense during the execution of the tactic. The combination of the two can construct a complete tactical execution simulation system, enabling the tactical demonstration to not only be limited to the tactical execution of the offensive side, but also dynamically adapt to the changes of the defensive side, improving the authenticity of the tactical simulation.

[0065] When training the tactical process model, the tactical simulation result mainly includes four aspects: the position change process of the tactical execution player, the position change process of the defensive player, the simulation process of the basketball transfer path, and the predicted value of the final shooting percentage. First, the position change process of the tactical execution player is used to track the running trajectory of the offensive player during the execution of the tactic. For different tactics, such as pick-and-roll, cut, and off-ball screen shooting, the movement patterns of the players are different, and this process can accurately simulate the real running trajectories of the players under different tactical backgrounds. The probability distribution of the player's future position is calculated through the hidden Markov model.

[0066] The position change process of the defensive player is used to simulate the tactical adjustment strategy of the defense. The movement of the defensive player is not only affected by their own tactical arrangements, but also closely related to the behavior of the offensive player. The model predicts the position adjustment of the defense by calculating the state transition probability of the defensive player. For example, when defending a pick-and-roll tactic, the defensive player may choose to switch defenses, go around the screen, or delay the offense, and these changes directly affect the effectiveness of the tactical execution. The defensive influence factor has a key impact on the player's shooting decision and passing path.

[0067] The simulation process of the basketball transfer path is used to track the passing method of the ball during the execution of the tactic. Under different tactics, the passing patterns of the basketball may include ordinary passes, bounce passes, and lob passes. The choice of passing method directly affects the success rate of the tactical execution. The model dynamically calculates the best passing path by analyzing the positions of the players, the blocking situation of the defensive players, and the passing success rate, improving the practical value of the tactical simulation. The predicted value of the shooting percentage is used to evaluate the final scoring possibility of the tactic. This prediction is based on various factors such as the player's historical shooting data, defensive pressure, and game rhythm, and calculates the shooting success probability through logistic regression.

[0068] Based on the solution of the present invention, by integrating the movement model of offensive players and the interference model of defensive players, the tactical simulation system has higher dynamic adaptability and realism. Compared with the traditional static tactical demonstration method, this method can not only track the movement paths of players, but also simulate the confrontation process between offense and defense, and finally calculate the shooting success rate of tactical execution, providing accurate data support for tactical optimization and game strategies.

[0069] Step S50: Perform visualization processing on the tactical simulation result and push the visualization result to the user side.

[0070] Specifically, based on the position change process of the tactical execution players and the position change process of the defensive players, a dynamic tactical demonstration screen is generated; real-time tactical data annotation is performed on the dynamic tactical demonstration screen; wherein, the tactical data includes any one or more of the player running route, passing trajectory, and shooting hit rate at the real-time position; the annotated dynamic tactical demonstration screen is pushed to the user side.

[0071] In the embodiment of the present invention, in the visualization processing, based on the position change process of the tactical execution players and the position change process of the defensive players, a dynamic tactical demonstration screen is generated to display the movement trajectories of the players, defensive reactions, and the passing paths of the basketball during the tactical execution process. Compared with the traditional static tactical diagrams, this dynamic demonstration screen can more clearly present the real-time standing position changes of the players during the tactical execution process, enabling coaches and players to more intuitively understand the tactical operation logic.

[0072] After generating the dynamic tactical demonstration screen, in order to further enhance the accuracy of tactical analysis, real-time tactical data annotation is performed on the demonstration screen, enabling users to not only observe the overall process of tactical execution, but also obtain data information on key tactical indicators. The content of tactical data annotation includes but is not limited to the player running route, passing trajectory, and shooting hit rate at the real-time position. Among them, the annotation of the player running route can use curves of different colors to represent the movement trajectories of the players and dynamically update the position changes of the players in combination with the time axis. The passing trajectory can be displayed by arrows or animation effects to show the passing methods of the basketball between different players, making the passing choices of the tactics clearer. For the shooting hit rate at the real-time position, based on the current standing position of the player, the shooting success rate at this position can be calculated and displayed, and the numerical value can be intuitively presented in the form of floating labels or heat maps.

[0073] Furthermore, after the annotation is completed, the generated dynamic tactical demonstration screen is pushed to the user side, supporting interactive display on the tactical analysis terminal, mobile device or tactical coaching board. Users can customize the tactical analysis method by adjusting the perspective, replaying key tactical segments, viewing different tactical data layers (such as only showing running routes or only viewing shooting heat maps), etc., to improve the efficiency of tactical teaching and training. In addition, the visualization interface can also combine 3D modeling technology to make the movements of players, shooting and defensive actions more realistic, thereby enhancing the immersion and intuitiveness of tactical analysis.

[0074] Based on the solution of the present invention, by combining dynamic tactical demonstration and data annotation, tactical visualization not only stays at the level of the movement trajectories of players, but also can combine key tactical indicators such as shooting percentage and passing options to achieve more refined tactical analysis. Compared with traditional tactical boards or static tactical diagrams, this method can display the dynamic changes during the execution of tactics in real time and optimize the tactical demonstration in a data-driven manner, providing more scientific tactical understanding and decision-making support for coaches and players.

[0075] Figure 2 It is the system structure diagram of the basketball tactical demonstration system provided by an embodiment of the present invention. As Figure 2 shown, the embodiment of the present invention provides a basketball tactical demonstration system, and the system includes: an analysis unit, configured to identify the corresponding tactical serial number and the player information of the players executing the tactic in response to the tactical demonstration request signal of the user; a first model calling unit, configured to call the on-court movement model of the corresponding player based on the player information of the players executing the tactic as the first model; a second model calling unit, configured to call the defensive model of the corresponding team based on the opponent information on the current court as the second model; a simulation unit, configured to perform training on the tactical process model under the corresponding tactical serial number based on the first model and the second model to obtain the corresponding tactical simulation result; and a visualization unit, configured to perform visualization processing on the tactical simulation result and push the visualization result to the user side.

[0076] The embodiment of the present invention also provides a computer-readable storage medium, on which instructions are stored, and when the instructions run on a computer, the computer is made to execute the above-mentioned basketball tactical demonstration method.

[0077] Those skilled in the art can understand that all or part of the steps in the methods for implementing the above embodiments can be completed by instructing relevant hardware through a program. This program is stored in a storage medium and includes several instructions to enable a single-chip microcomputer, a chip, or a processor to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0078] The optional embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above embodiments. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention. Additionally, it should be noted that, among the various specific technical features described in the above specific embodiments, they can be combined in any suitable manner without conflict. To avoid unnecessary repetition, the embodiments of the present invention will not separately describe various possible combination methods.

[0079] In addition, any combination can be made among various different embodiments of the present invention as long as it does not violate the idea of the embodiments of the present invention, and it should also be regarded as the content disclosed in the embodiments of the present invention.

Claims

1. A basketball tactical demonstration method, characterized in that: The method comprises: In response to a tactical demonstration request signal from a user, identifying a corresponding tactical sequence number and information of a player executing the tactic; Based on the player information of the executed tactics, the on-field movement model of the corresponding player is called as the first model; wherein, The on-court movement model is obtained by training a training sample composed of on-court movement history data of the corresponding player; wherein the on-court movement history data includes: any one or more of the movement speed of the corresponding player in an unguarded state, the movement speed bypassing a pick-and-roll, the actual position on the court at each moment, and the shooting percentage data of each point under different defensive pressure levels; The training rules of the on-court movement model include: performing data preprocessing on the on-court movement history data of the target player; based on the on-court movement history data after data preprocessing, taking the movement speed of the corresponding player in an undefended state and the movement speed bypassing the pick-and-roll as the first training sample and the actual on-court position of the corresponding player at each moment; taking the shooting hit rate data of each point of the corresponding player under different defensive pressure levels as the second training sample; performing model training based on the hidden Markov model algorithm and the first training sample to obtain the on-court position prediction model of the corresponding player; performing model training based on the logistic regression algorithm and the second training sample to obtain the shooting hit rate prediction model of the corresponding player; performing the coupling of the on-court position prediction model and the shooting hit rate prediction model to obtain the on-court movement model of the corresponding player; Based on the opponent information on the current field, the corresponding team's defense model is called as the second model; wherein, The training rules corresponding to the team's defensive model include: Predict the future position distribution of defensive players based on the hidden Markov model to calculate the state transition probability of the defensive position based on the individual defensive ability, defensive tactical strategy and offensive player behavior of the defensive players; The defensive influence factor is calculated based on the state transition probability of the defensive position; among them, The defensive influencing factors include the interference degree of the defensive distance on the shooting accuracy, the probability of the defensive player blocking the shot, and the influence degree of the steal on the passing success rate; Taking the defensive influencing factor as the influencing factor of the tactical execution process, simulating the decision-making changes of offensive players under different defensive modes, and obtaining the initial model; Based on historical game data, the position model and defense influencing factor of each defensive player of the corresponding team in the initial model are adjusted, and the model parameters are adjusted in combination with the actual game situation under the corresponding historical data to obtain the defense model of the corresponding team; Based on the first model and the second model, the tactical process model training under the corresponding tactical sequence number is performed to obtain the corresponding tactical simulation result; wherein, The tactical simulation results include: The position change process of the tactical execution player, the position change process of the defensive player, the simulation process of the basketball transfer path, and the final shooting percentage prediction value; Visualization processing is performed on the tactical simulation results, and the visualization results are pushed to the user end.

2. The method according to claim 1, characterized in that The performing of model training based on the hidden Markov model algorithm and the first training sample to obtain a corresponding player's on-field position prediction model includes: The tactical states under different tactical numbers are used as hidden variables of the hidden Markov model algorithm, and the actual positions of the corresponding players at each moment are used as observation data of the hidden Markov model algorithm to obtain the corresponding first training samples; Setting a state transfer matrix, an observation probability matrix and an initial state distribution of a hidden Markov model based on the first training sample; In the iterative process, the expected value of each round of iteration is calculated based on the forward-backward algorithm, and the state transfer matrix, the observation probability matrix and the initial state distribution are adjusted by maximizing the likelihood function until the convergence condition is met to obtain the optimal state transfer matrix and observation probability matrix; Based on the optimal state transfer matrix and observation probability matrix, a corresponding hidden Markov model is constructed as the on-field position prediction model of the corresponding player.

3. The method according to claim 1, characterized in that The method of performing model training based on the logistic regression algorithm and the second training sample to obtain a shooting percentage prediction model for the corresponding player includes: The shooting position, defensive pressure level, player shooting type and game background information of each data in the second training sample are annotated to obtain the annotated second training sample; wherein, The defensive pressure level is determined based on the distance between the defensive player and the current player and the maximum defensive height of the defensive player; The game background information is the total playing time of the current player in this game; Using the annotated second training sample as an input feature of the logistic regression model to perform model training; During the model training process, the loss function is calculated based on the maximum likelihood estimation, and the parameters are adjusted by gradient descent until the convergence conditions are met to obtain the corresponding player's shooting percentage prediction model.

4. The method according to claim 1, characterized in that Performing visualization processing on the tactical simulation results and pushing the visualization results to the user end includes: Generate dynamic tactical demonstration screens based on the position change process of tactical execution players and defensive players; Perform real-time annotation of tactical data in the dynamic tactical demonstration screen; wherein, The tactical data includes: Any one or more of the player's running route, passing trajectory and shooting percentage at real-time position; Push the annotated dynamic tactical demonstration screen to the user end.

5. A basketball tactical demonstration system, characterized in that: The system comprises: A parsing unit, for responding to a tactical demonstration request signal from a user, identifying a corresponding tactical sequence number and information about players executing the tactic; The first model calling unit is used to call the on-field movement model of the corresponding player based on the player information executing the tactics as the first model; wherein, The on-court movement model is obtained by training a training sample composed of on-court movement history data of the corresponding player; wherein the on-court movement history data includes: any one or more of the movement speed of the corresponding player in an unguarded state, the movement speed bypassing a pick-and-roll, the actual position on the court at each moment, and the shooting percentage data of each point under different defensive pressure levels; The training rules of the on-court movement model include: performing data preprocessing on the on-court movement history data of the target player; based on the on-court movement history data after data preprocessing, taking the movement speed of the corresponding player in an undefended state and the movement speed bypassing the pick-and-roll as the first training sample and the actual on-court position of the corresponding player at each moment; taking the shooting hit rate data of each point of the corresponding player under different defensive pressure levels as the second training sample; performing model training based on the hidden Markov model algorithm and the first training sample to obtain the on-court position prediction model of the corresponding player; performing model training based on the logistic regression algorithm and the second training sample to obtain the shooting hit rate prediction model of the corresponding player; performing the coupling of the on-court position prediction model and the shooting hit rate prediction model to obtain the on-court movement model of the corresponding player; The second model calling unit is used to call the corresponding team's defense model as the second model based on the opponent information on the current field; wherein, The training rules corresponding to the team's defensive model include: Predict the future position distribution of defensive players based on the hidden Markov model to calculate the state transition probability of the defensive position based on the individual defensive ability, defensive tactical strategy and offensive player behavior of the defensive players; The defensive influence factor is calculated based on the state transition probability of the defensive position; among them, The defensive influencing factors include the interference degree of the defensive distance on the shooting accuracy, the probability of the defensive player blocking the shot, and the influence degree of the steal on the passing success rate; Taking the defensive influencing factor as the influencing factor of the tactical execution process, simulating the decision-making changes of offensive players under different defensive modes, and obtaining the initial model; Based on historical game data, the position model and defense influencing factor of each defensive player of the corresponding team in the initial model are adjusted, and the model parameters are adjusted in combination with the actual game situation under the corresponding historical data to obtain the defense model of the corresponding team; A simulation unit is used to perform tactical process model training under the corresponding tactical sequence number based on the first model and the second model to obtain a corresponding tactical simulation result; wherein, The tactical simulation results include: The position change process of the tactical execution player, the position change process of the defensive player, the simulation process of the basketball transfer path, and the final shooting percentage prediction value; The visualization unit is used to perform visualization processing on the tactical simulation results and push the visualization results to the user end.

6. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the basketball tactical demonstration method described in any one of claims 1-4.

Images