Basketball shooting training system and device based on intelligent image analysis

Abstract

According to the basketball shooting training system and device based on intelligent image analysis, the hand detection module, the gesture key point detection module, the hand shape analysis module, the shooting track simulation module, the intelligent feedback generation module, the data management module, the user interaction interface module and the like are integrated, and the system analyzes hand actions of a trainer through videos, detects joint points and compares the hand actions with standard gestures; according to the device, the hand shape difference is evaluated, a physical simulation algorithm is utilized, a basketball flight path is simulated according to a hand shape analysis result, hit rate prediction is provided, an intelligent feedback module generates personalized training suggestions according to a simulation result and historical data, and the whole device helps trainees to optimize shooting skills and improve the training effect through real-time data and visual feedback.

Description

Technical Field

[0001] The present invention relates to the technical field of sports training, and more specifically, to a basketball shooting training system and device based on intelligent image analysis. Background Art

[0002] With the popularity and increasing attention of basketball, more and more people are joining in basketball training. Shooting is one of the core techniques of basketball, and its training methods and effects are crucial to the improvement of players' skills. However, traditional shooting training methods have many shortcomings, such as one-on-one guidance from professional coaches is expensive and time-inefficient, watching video tutorials lacks training feedback and recognition of action differences, and the use of equipment such as sensors is inconvenient and has a low penetration rate. Therefore, there is an urgent need for a training system that can intelligently analyze shooting actions and provide real-time feedback.

[0003] The existing basketball training system relies on the subjective judgment of coaches and the self-perception of athletes, lacks precise data support and objective analysis, resulting in limited shooting training effects. Summary of the invention

[0004] In order to overcome the problem that the existing basketball training system relies on the subjective judgment of the coach and the self-perception of the athletes, lacks accurate data support and objective analysis, resulting in limited shooting training effects, the present invention designs a basketball shooting training system and device based on intelligent image analysis that can effectively solve the above technical problems.

[0005] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0006] A basketball shooting training system based on intelligent image analysis, comprising: a hand detection module, a gesture key point detection module, a hand shape analysis module, a shooting trajectory simulation module, an intelligent feedback generation module, a data management and analysis module and a user interaction interface;

[0007] A hand detection module is used to detect and locate the detection frame of the trainee's hand in each frame of the video recording the training state;

[0008] The gesture key point detection module is used to process the image within the hand position detection frame detected by the hand detection module to obtain the hand joint points;

[0009] The hand shape analysis module is used to analyze the gesture key points obtained by the gesture key point detection module, and compare and analyze the gesture key point information with the preset standard gesture;

[0010] Shooting trajectory simulation module, used to simulate the flight trajectory of the basketball according to the hand shape analysis results;

[0011] Intelligent feedback generation module, used to generate feedback and improvement suggestions during shooting training;

[0012] Data management and analysis module, used to record and store shooting training data;

[0013] User interaction interface, used to display real-time shooting training data and historical shooting training data.

[0014] Preferably, the hand detection module further comprises the following steps:

[0015] Receive a shooting training video and decompose it into continuous frame images;

[0016] Apply skin color detection algorithm to each frame of image to preliminarily filter out the area containing the hand;

[0017] Apply shape analysis and machine learning models to accurately detect and locate the outline of the hand in the above-screened area to form a hand detection frame;

[0018] Output the hand detection frame coordinates in each frame image to the gesture key point detection module.

[0019] Preferably, the gesture key point detection module further comprises the following steps:

[0020] Receive the detection frame image of the hand, and pre-process the image within the detection frame;

[0021] Apply deep learning models to extract features from hand images;

[0022] Based on the extracted features, the hand joints are detected and located using regression algorithms or heat map methods;

[0023] Output the coordinate information of gesture key points to the hand shape analysis module.

[0024] Preferably, the hand shape analysis module further comprises the following steps:

[0025] Receive gesture key point coordinate information;

[0026] Calculate hand posture parameters based on key point coordinates, such as finger bending angle, palm orientation, etc.

[0027] Compare the calculated hand gesture parameters with a preset standard shooting gesture database, which contains key point coordinates and gesture parameters of multiple standard shooting gestures;

[0028] Evaluate the differences between the trainee's shooting hand shape and the standard hand shape, including the degree and type of posture deviation;

[0029] Output the analysis results to the shooting trajectory simulation module and the intelligent feedback generation module.

[0030] Preferably, the shooting trajectory simulation module further comprises the following steps:

[0031] Receive hand shape analysis results and the trainee's position information at the moment of shooting;

[0032] Apply physical simulation algorithms to simulate the flight trajectory of the basketball based on the hand shape analysis results;

[0033] Predict where the basketball will land and compare it with the target basket position to calculate the shooting percentage;

[0034] Output the simulated shooting trajectory and predicted hit rate to the intelligent feedback generation module.

[0035] Preferably, the intelligent feedback generation module further includes the following technical steps:

[0036] Receive shooting trajectory simulation results, hand shape analysis results and historical training data;

[0037] Analyze the common mistakes that trainees make during shooting;

[0038] Generate personalized training suggestions based on the analysis results, including adjusting shooting posture, strengthening strength training or improving shooting rhythm;

[0039] The feedback suggestions are presented to the trainer in the form of text, charts or video presentations through the user interaction interface.

[0040] A basketball shooting training device based on intelligent image analysis is implemented by the above-mentioned basketball shooting training system based on intelligent image analysis.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows: the hand detection module of the present invention receives the training video and decomposes it into frame images, applies the skin color detection algorithm to preliminarily screen out the hand area, and then uses shape analysis and machine learning models to accurately locate the outline of the hand, thereby reducing the reliance of coaches and athletes on manual detection of hand positions and improving the accuracy and consistency of detection; the gesture key point detection module preprocesses the image within the hand detection frame and applies the deep learning model to extract features, and then uses the regression algorithm or the heat map method to detect and locate the hand joints. This method provides accurate joint point data of the hand posture and provides detailed key point information for further analysis, surpassing the subjective judgment of coaches and athletes; the hand shape analysis module receives the gesture key point coordinate information, calculates the hand posture parameters, and compares them with the preset standard shooting gesture database. This comparison provides an objective analysis to evaluate the difference between the trainee's hand shape and the standard hand shape, including the degree and type of posture deviation, thereby providing athletes with specific improvement directions; the shooting trajectory simulation module calculates the hand shape according to the hand shape. The analysis results and the trainee's position information at the moment of shooting are used to simulate the flight trajectory of the basketball and predict the landing point of the basketball. This simulation provides a prediction of the shooting effect and helps athletes understand the impact of different hand shapes on shooting results, which is not available in traditional training systems. The intelligent feedback generation module receives the shooting trajectory simulation results, hand shape analysis results and historical shooting training data, analyzes the trainee's common mistakes and room for improvement, and generates personalized training suggestions. This feedback mechanism provides customized improvement suggestions to help athletes improve their shooting skills in a targeted manner. The data management and analysis module is responsible for recording and storing shooting training data, and provides coaches and athletes with detailed training reports and analysis. This data management capability ensures the traceability of training progress and effects, and provides athletes with a basis for continuous improvement. The user interaction interface displays real-time shooting training data and historical shooting training data, and provides feedback in the form of text, charts or video presentations. This intuitive display method enables athletes to easily understand their shooting performance and improvement directions. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived based on the provided drawings without paying any creative work.

[0043] Figure 1 This is a structural diagram of a basketball shooting training system based on intelligent image analysis. DETAILED DESCRIPTION

[0044] The drawings are for illustrative purposes only and should not be construed as limiting the present patent;

[0045] In order to better illustrate the present embodiment, some parts in the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product;

[0046] It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0047] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0048] Example

[0049] A basketball shooting training system based on intelligent image analysis, such as Figure 1 As shown, it includes: hand detection module, gesture key point detection module, hand shape analysis module, shooting trajectory simulation module, intelligent feedback generation module, data management and analysis module and user interaction interface;

[0050] A hand detection module is used to detect and locate the detection frame of the trainee's hand in each frame of the video recording the training state;

[0051] The gesture key point detection module is used to process the image within the hand position detection frame detected by the hand detection module to obtain the hand joint points;

[0052] The hand shape analysis module is used to analyze the gesture key points obtained by the gesture key point detection module, and compare and analyze the gesture key point information with the preset standard gesture;

[0053] Shooting trajectory simulation module, used to simulate the flight trajectory of the basketball according to the hand shape analysis results;

[0054] Intelligent feedback generation module, used to generate personalized feedback and improvement suggestions;

[0055] Data management and analysis module, used to record and store training data;

[0056] User interaction interface for displaying real-time data and historical records.

[0057] The hand detection module further comprises the following steps:

[0058] Receive a training video and decompose it into consecutive frame images;

[0059] Apply skin color detection algorithm to each frame of image to preliminarily filter out the area containing the hand; such as threshold skin color recognition based on RGB color space:

[0060] In the RGB model, skin color recognition usually requires complex constraints because skin color is greatly affected by brightness.

[0061] For light skin color, the range of skin color under the RGB model basically satisfies the following constraints: (R, G, B)>(95, 40, 20) and max(R, G, B)-min(R, G, B)>15 and |RG|>15 and R>G and R>B.

[0062] For dark skin, the recognition conditions of dark skin color may be R>220, G>210, B>170, |RG|<=15, R>BG>B.

[0063] Among them, (R,G,B) represents the values ​​of red (R), green (G), and blue (B);

[0064] Apply shape analysis and machine learning models to accurately detect and locate the outline of the hand in the above-screened area to form a hand detection frame;

[0065] Output the hand detection frame coordinates in each frame image to the gesture key point detection module.

[0066] The gesture key point detection module further comprises the following steps:

[0067] Receive the hand detection frame image and pre-process the image in the detection frame, including image enhancement and noise removal;

[0068] Apply deep learning models, such as convolutional neural networks (CNNs), to extract features from hand images;

[0069] Based on the extracted features, the hand joints are detected and located using regression algorithms or heat map methods, including but not limited to finger joints, wrists, etc.;

[0070] The specific steps of the regression algorithm are as follows: Step 1: Input the extracted feature map into the trained regression model.

[0071] Step 2: The regression model outputs the predicted coordinate values ​​of each joint point.

[0072] Step 3: Post-process the predicted joint point coordinates, such as using non-maximum suppression (NMS) to remove overlapping or redundant prediction points.

[0073] Step 4: Calibrate the predicted joint point coordinates to compensate for any deformation or scaling during image preprocessing.

[0074] The coordinates of the joint points predicted by the regression model are taken as output, which correspond to the positions of finger joints, wrists, etc.

[0075] The steps of heat map detection are:

[0076] The CNN model will learn to identify the locations of hand joints from the image and generate a corresponding heat map.

[0077] Heatmap generation,The model infers the input hand image and generates a probability density function,for each hand joint, which is expressed as a heatmap.,Each joint point has a corresponding heatmap channel.,The model locates the joint point by finding the highest response area,on the heatmap.

[0078] Keypoint localization,The location of each joint point is extracted from the heatmap. This is usually achieved by applying the argmax function on the corresponding heatmap to find the exact location of each joint.

[0079] Post-processing,The joint point positions output by the model are post-processed, such as removing outliers,,smoothing, etc., to improve the accuracy of positioning.

[0080] Evaluation and optimization: Use the validation set to evaluate the hand joint detection model and calculate the error of key point positioning, such as the average Euclidean distance, etc. Optimize the model based on the evaluation results, such as adjusting the network structure, loss function or training strategy, etc.

[0081] Output the coordinate information of gesture key points to the hand shape analysis module.

[0082] The hand shape analysis module further comprises the following steps:

[0083] Receive gesture key point coordinate information;

[0084] Calculate hand posture parameters based on key point coordinates, such as finger bending angle, palm orientation, etc.

[0085] Get the key point coordinates and obtain the hand joint point coordinates from the gesture key point detection module. These coordinates usually include key positions such as the palm center, fingertips, and knuckles.

[0086] Define the reference coordinate system:

[0087] Establish a three-dimensional or two-dimensional reference coordinate system to uniformly measure hand posture; second, calculate the finger bending angle

[0088] Determine the finger joints by identifying and labeling the proximal knuckles (the joints closest to the palm), middle knuckles (if present), and distal knuckles (fingertips) of each finger.

[0089] Calculate the joint vectors. For each finger, calculate the vectors between adjacent joint points. For example, for the proximal and middle finger joints, calculate the vector from the proximal finger joint to the middle finger joint; for the middle and distal finger joints, calculate the corresponding vectors.

[0090] To calculate the bending angle, use the vector cross product and dot product to calculate the angle between two adjacent joint vectors. This angle represents the bending angle of the finger at that joint.

[0091] Formula: cosθ = (A·B) / (|A|*|B|), where A and B are adjacent joint vectors and θ is the angle between them. The value of θ is then calculated using the arccosine function (arccos).

[0092] Calculate palm orientation

[0093] Determine the palm key points, identify and mark the palm center point and wrist point (or the point where the palm connects to the arm).

[0094] The palm plane is calculated by using the palm center point and the proximal knuckle points of at least three fingers (usually the proximal knuckles of the thumb, index finger and middle finger are selected) to define a plane, which can be regarded as the plane of the palm.

[0095] From these four points, a normal vector can be calculated, which is perpendicular to the plane of the palm.

[0096] Determine the palm orientation. The palm orientation can be determined by the normal vector of the palm plane. In three-dimensional space, this normal vector can be expressed as a three-dimensional vector, whose direction indicates the orientation of the palm.

[0097] In a two-dimensional image, it can be simplified to calculate the angle between the palm plane and the image plane, so as to determine whether the palm is facing the audience or facing away from the audience.

[0098] Compare the calculated hand gesture parameters with a preset standard shooting gesture database, which contains key point coordinates and gesture parameters of multiple standard shooting gestures;

[0099] Evaluate the differences between the trainee's shooting hand shape and the standard hand shape, including the degree and type of posture deviation;

[0100] Output the analysis results to the shooting trajectory simulation module and the intelligent feedback generation module.

[0101] The shooting trajectory simulation module further includes the following steps:

[0102] Receive hand shape analysis results and the trainee's position information at the moment of shooting;

[0103] Apply physical simulation algorithms, such as parabolic equations or more complex aerodynamic models, to simulate the flight trajectory of the basketball based on hand shape analysis results, such as shooting angle and strength;

[0104] Data collection, through the hand shape analysis module to obtain key parameters such as initial speed, angle, rotation speed, etc. during shooting.

[0105] Model selection, choosing the appropriate physics simulation algorithm. This can be based on parabolic equations from classical mechanics or a more complex aerodynamic model that takes into account air resistance and the Magnus effect.

[0106] Establish the dynamic equation. For the case of no air resistance, use the parabolic equation: X = V0Tcos(θ)y = V0Tsin(θ)-1 / 2gt^2+h0;

[0107] Among them, V0 is the initial velocity, θ is the shooting angle, g is the acceleration of gravity, h0 is the shooting height, and t is the time.

[0108] For cases where air resistance is taken into account, more complex dynamic equations are used and may need to be solved by numerical methods.

[0109] Numerical solution: Use numerical methods (such as Euler method, Runge-Kutta method, etc.) to solve the dynamic equations and obtain the flight trajectory of the basketball.

[0110] Simulate the trajectory, based on the numerical solution results, simulate the flight trajectory of the basketball. Graphical tools can be used to visualize the trajectory for a more intuitive display.

[0111] Result Analysis,Analyze the simulation results, predict the landing point of the basketball, and compare it with the target basket position to calculate the shooting percentage.

[0112] Predict where the basketball will land and compare it with the target basket position to calculate the shooting percentage;

[0113] Output the simulated shooting trajectory and predicted hit rate to the intelligent feedback generation module.

[0114] The intelligent feedback generation module further includes the following technical steps:

[0115] Receive shooting trajectory simulation results, hand shape analysis results and historical training data;

[0116] Analyze the common mistakes that trainees make during shooting;

[0117] Generate personalized training suggestions based on the analysis results, including adjusting shooting posture, strengthening strength training or improving shooting rhythm;

[0118] The feedback suggestions are presented to the trainer in the form of text, charts or video presentations through the user interaction interface.

[0119] A basketball shooting training device based on intelligent image analysis is realized by the above-mentioned basketball shooting training system based on intelligent image analysis.

[0120] The same or similar reference numerals correspond to the same or similar components;

[0121] The terms used in the drawings to describe positional relationships are only used for illustrative purposes and should not be construed as limiting this patent;

[0122] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not limitations on the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made on the basis of the above description. It is not necessary and impossible to list all the implementation methods here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A basketball shooting training system based on intelligent image analysis, characterized in that: include: Hand detection module, gesture key point detection module, hand shape analysis module, shooting trajectory simulation module, intelligent feedback generation module, data management and analysis module and user interaction interface; A hand detection module is used to detect and locate the detection frame of the trainee's hand in each frame of the video recording the training state; The gesture key point detection module is used to process the image within the hand position detection frame detected by the hand detection module to obtain the hand joint points; The hand shape analysis module is used to analyze the gesture key points obtained by the gesture key point detection module, and compare and analyze the gesture key point information with the preset standard gesture; Shooting trajectory simulation module, used to simulate the flight trajectory of the basketball according to the hand shape analysis results; Intelligent feedback generation module, used to generate feedback and improvement suggestions during shooting training; Data management and analysis module, used to record and store shooting training data; User interaction interface, used to display real-time shooting training data and historical shooting training data.

2. The basketball shooting training system based on intelligent image analysis according to claim 1, characterized in that: The detection frame for detecting and locating the trainer's hand in each frame of the video for recording the training state further comprises the following steps: Receive a shooting training video and decompose it into continuous frame images; Apply skin color detection algorithm to each frame of image to preliminarily filter out the area containing the hand; Apply shape analysis and machine learning models to accurately detect and locate the outline of the hand in the above-screened area to form a hand detection frame; Output the hand detection frame coordinates in each frame image to the gesture key point detection module.

3. The basketball shooting training system based on intelligent image analysis according to claim 1, characterized in that: The method for processing the image in the hand position detection frame detected by the hand detection module to obtain the hand joint points further includes the following steps: Receive the detection frame image of the hand, and pre-process the image within the detection frame; Apply deep learning models to extract features from hand images; Based on the extracted features, the hand joints are detected and located using regression algorithms or heat map methods; Output the coordinate information of gesture key points to the hand shape analysis module.

4. The basketball shooting training system based on intelligent image analysis according to claim 1, characterized in that: The method for performing action analysis based on the gesture key points obtained by the gesture key point detection module and performing comparative analysis with the preset standard gesture key point information further includes the following steps: Receive gesture key point coordinate information; Calculate hand posture parameters based on key point coordinates; Compare the calculated hand gesture parameters with a preset standard shooting gesture database, which contains key point coordinates and gesture parameters of multiple standard shooting gestures; Evaluate the differences between the trainee's shooting hand shape and the standard hand shape, including the degree and type of posture deviation; Output the analysis results to the shooting trajectory simulation module and the intelligent feedback generation module.

5. The basketball shooting training system based on intelligent image analysis according to claim 1, characterized in that: The method for simulating the flight trajectory of a basketball according to the hand shape analysis result further comprises the following steps: Receive hand shape analysis results and the trainee's position information at the moment of shooting; Apply physical simulation algorithms to simulate the flight trajectory of the basketball based on the hand shape analysis results; Predict where the basketball will land and compare it with the target basket position to calculate the shooting percentage; Output the simulated shooting trajectory and predicted hit rate to the intelligent feedback generation module.

6. The basketball shooting training system based on intelligent image analysis according to claim 1, characterized in that: The method for generating feedback and improvement suggestions for the shooting training process further includes the following steps: Receive shooting trajectory simulation results, hand shape analysis results, and historical shooting training data; Analyze the common mistakes that trainees make during shooting; Generate personalized training suggestions based on the analysis results, including adjusting shooting posture, strengthening strength training or improving shooting rhythm; The feedback suggestions are presented to the trainer in the form of text, graphics or video presentation through the user interaction interface.

7. A basketball shooting training device based on intelligent image analysis, characterized in that: This is achieved through a basketball shooting training system based on intelligent image analysis as described in claims 1-6.

Images