Football examination system based on AI vision

Abstract

The invention relates to the technical field of AI vision, and provides a football examination system based on AI vision, an image acquisition module is used for acquiring image information in an examination process, a three-dimensional coordinate module is connected with the image acquisition module to establish a three-dimensional coordinate system, and three-dimensional coordinates of each point in an image are calculated; the recognition module is connected with the three-dimensional coordinate module and calculates the motion trails of the examinee and the football, the judgment module is connected with the recognition module and judges foul and goal of the examinee according to preset rules, the analysis module is connected with the judgment module and calculates examination scores according to the judgment result of the judgment module and generates examinee training suggestions, and the data management module is connected with the analysis module. Storing the examination data, and generating an examinee score report and an examinee training suggestion report; through the recognition module, the judgment module and the analysis module, the examination efficiency is improved, the examination process is more scientific and normative, examinees can clearly know own examination performance and defects, and a clear direction is provided for subsequent training.

Description

Technical Field

[0001] The present invention relates to the field of AI vision technology, in particular to a football examination system based on AI vision. Background Art

[0002] In recent years, AI vision technology has achieved rapid development, and its application has been continuously expanded and deepened. In the field of physical education and examinations, using AI vision technology to improve efficiency and quality has become a new trend. As one of the most popular sports in the world, football has strict requirements for talent training and selection. However, traditional football examinations rely on manual referees and manual records, which are inefficient and easily interfered by human factors, and the accuracy and fairness of the test results are questionable. At the same time, with the development of football technology and the increasing complexity of competition rules, as well as the increasing number of people participating in football examinations, the limitations of human attention and judgment in traditional examination methods make it difficult to accurately analyze and judge a large amount of test data in a short period of time, which also brings huge challenges to the organization and management of the examination.

[0003] The current judging of football tests is often in the following form: there is a referee at the starting point and the end point, who determines the final score of the tester based on the tester's comprehensive performance during the test, such as judging whether the ball goes around the pole or a goal is scored, and manually times the time with a stopwatch.

[0004] The existing background technology still has the following shortcomings:

[0005] 1. In the existing soccer ball passing the line or soccer ball plus goal test, the manual judgment method is prone to errors in the test results due to differences in individual judgment standards, making it difficult to unify the test standards. In addition, the test mode is inefficient, and manual referees need to concentrate on tracking the movements and performance of each candidate. When a large number of candidates participate in the test, they are prone to fatigue and lack of concentration, which affects the timeliness and accuracy of the judgment.

[0006] 2. The way the infrared ray beam-forming device judges the violation of the pole and the goal of the football is heavily dependent on the infinite sensing logic of the hardware, which is easily affected by environmental factors. For example, in windy weather, the infrared ray may be deviated or interfered; in strong light conditions, there may be light obstruction or reflection problems, resulting in sensing errors, and thus unable to accurately judge the pole and goal situations in the test.

[0007] 3. The current examination system lacks a comprehensive analysis of the examination process, and cannot provide detailed action standards and process feedback to candidates, nor can it provide reasonable improvement plans and methods to help examinees improve their test-taking skills and scores. After the examination, candidates can only get a score, but do not know where they have problems and how to improve, which is not conducive to the technical improvement and long-term development of candidates. Summary of the invention

[0008] In order to solve the above technical problems, the present invention provides a football examination system based on AI vision to solve the problems in the existing background technology.

[0009] The technical solution of the present invention is as follows:

[0010] A football examination system based on AI vision, comprising:

[0011] An image acquisition module, which is used to acquire image information during the examination;

[0012] A three-dimensional coordinate module, which is communicatively connected to the image acquisition module, establishes a three-dimensional coordinate system based on the image information acquired by the image acquisition module, and calculates the three-dimensional coordinates of each point in the image;

[0013] An identification module, the identification module being communicatively connected to the three-dimensional coordinate module and used for identifying and calculating the motion trajectory of the examinee and the football according to the three-dimensional coordinates of each point in the image calculated by the three-dimensional coordinate module;

[0014] A judgment module, the judgment module being in communication connection with the identification module, and used to judge whether the examinee has committed a foul in the process of dribbling the ball around the pole and whether the ball has successfully scored a goal according to preset rules and the movement trajectory of the examinee and the ball identified by the identification module;

[0015] An analysis module, the analysis module being communicatively connected to the judgment module, and configured to automatically calculate the test scores according to the judgment results of the judgment module, and generate training suggestions for the examinees;

[0016] The data management module is connected to the analysis module in communication, stores the test data, and generates a test taker's score report and a test taker's training suggestion report.

[0017] Preferably, the image acquisition module includes a camera angle adjustment unit and a camera parameter adjustment unit;

[0018] The camera angle adjustment unit is used to allocate the shooting angle of each camera according to the layout of the examination venue and the position distribution of the cameras;

[0019] The camera parameter adjustment unit is used to adjust the camera parameters including but not limited to frame rate, resolution, exposure time, gain and white balance. For locations in the center of the venue, the frame rate and resolution are increased, and for cameras in the edge area of ​​the venue, the frame rate and resolution are reduced; when the ambient light is strong, the exposure time is automatically shortened, and when the ambient light is weak, the gain is increased and the exposure time is extended.

[0020] Preferably, the three-dimensional coordinate module uses the image information collected by the image acquisition module to calculate the three-dimensional coordinates of each point in the image by matching the parallax of corresponding points in different camera images;

[0021] First, establish an accurate three-dimensional coordinate system for the test site, set up the x-axis, y-axis and z-axis; confirm the coordinate points in the three-dimensional coordinate system through any two cameras in the camera module;

[0022] The optical center of the first camera is O1, and the intersection of the camera optical axis and the imaging plane, i.e. the principal point coordinates, is (x o1 ,y o1 ), the focal length is f1; the optical center of the second camera is O2, the intersection of the camera optical axis and the imaging plane, the principal point coordinates are (x o2 ,y o2 ), focal length is f2;

[0023] b is the distance between the optical centers O1 and O2 of the two cameras, that is, the baseline distance b;

[0024] For a point P in space, the horizontal and vertical coordinates of the imaging plane of the first camera are (x1, y1), and the coordinates of the imaging plane of the second camera are (x2, y2). The parallax is the difference between the horizontal coordinates of the same spatial point in the imaging planes of the two cameras, d = (x1-x2);

[0025] The coordinates of the spatial point P in the first camera coordinate system are (x p1 ,y p1 ,z p1 );

[0026] According to the principle of similar triangles and triangulation, we have the following formula:

[0027]

[0028] From formula (1), we can get:

[0029]

[0030] Substituting (3) into (2), we get:

[0031]

[0032]

[0033] Substituting equation (5) into equation (3) and equation (4) respectively, we can obtain x p1 and p1 According to this method, the coordinates of each point photographed by the camera in the three-dimensional coordinate system can be established.

[0034] Preferably, the recognition module includes a coordinate extraction unit, a human motion recognition unit and a football trajectory recognition unit;

[0035] The coordinate extraction unit is used to extract coordinate points of the basketball and the human body;

[0036] The human motion recognition unit calculates the human motion trajectory according to the coordinates of the human skeleton key points extracted by the coordinate extraction unit;

[0037] The football trajectory recognition unit determines the position change trajectory of the football by matching the similarity of the football boundary coordinate vectors in adjacent frame images and then calculating the position change vector.

[0038] Preferably, the coordinate extraction unit extracts the coordinates of the football and the human body;

[0039] The football boundary coordinates are extracted based on shape detection and color detection technology, and the coordinates of these football boundary points in the three-dimensional coordinate system are obtained to calculate the football's motion trajectory. At the kth frame, the three-dimensional coordinates of several points on the football boundary form a boundary coordinate vector

[0040] The human body is based on skeleton recognition technology to extract skeleton key points, and obtain the coordinates of these key points in the three-dimensional coordinate system to calculate the human body movement trajectory.

[0041] Preferably, the human motion recognition unit is based on the specified time points t1, t2, t3, ..., t n And the corresponding three-dimensional coordinates of the skeleton key points Calculate the displacement between adjacent time points in sequence:

[0042] For the time period [t1, t2], the displacement Δs 1,2 for:

[0043]

[0044] Where s represents the displacement, That is, it is expressed as a displacement change vector;

[0045] The direction vector is:

[0046]

[0047] Repeating formula (8) at each different adjacent time point can obtain the displacement between each two adjacent time points: and direction vector Δs 2,3 ; Where i and j represent two adjacent time points; represents the direction vector;

[0048] Starting point As the starting point, each point is connected in sequence according to the calculated direction vector and displacement to obtain the motion trajectory of each key point of the human body within the specified time, thereby obtaining the motion trajectory of the human body.

[0049] Preferably, the football trajectory recognition unit calculates the similarity of the football boundary coordinate vectors in two adjacent frames of images;

[0050]

[0051] in, is the inner product of two eigenvectors, are the moduli of the two feature vectors respectively; if the similarity of the feature vectors of two adjacent frames of football is greater than 0.9, they are considered to be matched;

[0052] After determining that the football in the adjacent frame images is a matching object, calculate the position change of the football:

[0053] The coordinates of the football in the kth frame image are (x k ,y k ,z k ), the coordinates of the football in the k+1th frame image are (x k91 ,y k91 ,z k91 ); the position change vector is:

[0054]

[0055] The matching calculation formula (9) and the position change formula (10) are calculated for all adjacent frames to obtain the position change trajectory of the football in the entire image sequence.

[0056] Preferably, the judgment module includes but is not limited to using the following rules to judge violations:

[0057] Judgment of a foul on the pole: When the distance between the human body and the pole is less than or equal to 0, or the trajectory of the ball deviates from the prescribed route around the pole, it is judged as a foul on the pole; when it is greater than 0 or the trajectory of the ball does not deviate from the route around the pole, it is judged as a foul on not going around the pole;

[0058] Handball foul judgment: When the coordinates of the key points of the hand skeleton are detected to coincide with the coordinates of the football boundary or the distance is less than zero, it is judged as a handball foul; when they do not coincide or the distance is greater than zero, it is judged as no handball foul;

[0059] Goal judgment: Through straight line detection, the position and direction of the goal line are determined. When the boundary coordinates of the football cross the goal line, it is judged as a goal. If it does not cross, it is judged as a goal not scored.

[0060] Preferably, the analysis module generates performance analysis and training suggestions;

[0061] The performance analysis includes but is not limited to the calculation of foul deduction points and goal bonus points;

[0062] The foul deduction calculation: according to the judgment result of the judgment module, for each foul behavior, points are deducted according to the preset deduction rules;

[0063] Calculation of goal bonus: If the candidate successfully shoots the football into the goal, he / she will be awarded corresponding bonus points according to the difficulty, distance and angle of the goal;

[0064] The training suggestion generates targeted training suggestions according to the determination result of the determination module.

[0065] Preferably, the data management module includes a storage unit and a display unit;

[0066] The storage unit stores the results calculated by each module of the system respectively;

[0067] The display unit is used to display the examinee's score report and examinee's training suggestion report generated according to the analysis result of the analysis module.

[0068] Compared with the prior art, the present invention has the following beneficial effects:

[0069] 1. The present invention uses the camera angle adjustment unit and the camera parameter adjustment unit in the image acquisition module to flexibly allocate shooting angles and intelligently adjust camera parameters according to the layout of the examination site and the distribution of camera positions to ensure the collection of high-quality image information. The three-dimensional coordinate module calculates the three-dimensional coordinates of each point in the image to achieve accurate spatial modeling and data acquisition of the examination scene, greatly improving the spatial perception ability and data accuracy of the examination scene.

[0070] 2. The present invention accurately extracts the coordinates of key points of the human skeleton and the coordinates of the football boundary through the coordinated work of the coordinate extraction unit, the human motion recognition unit and the football trajectory recognition unit in the recognition module, accurately calculates the human motion trajectory and the football motion trajectory, provides a more accurate and detailed basis, and helps to improve the fairness and objectivity of the test evaluation.

[0071] 3. The present invention cooperates with the judgment module, the analysis module and the data management module. The judgment module judges foul behaviors and goal situations according to preset rules. The analysis module automatically calculates the test scores and generates training suggestions based on this. The data management module stores the relevant data and generates reports, making the entire test process more scientific and standardized. Candidates can clearly understand their test performance and shortcomings, providing a clear direction for subsequent training. It also facilitates the test management department to uniformly manage and analyze the test data. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 It is a work flow chart of the present invention.

[0073] Figure 2 It is a schematic diagram of the examination site of the present invention.

[0074] Figure 3 It is a recognition module recognition workflow diagram of the present invention. DETAILED DESCRIPTION

[0075] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0076] The present invention provides a football examination system based on AI vision, comprising:

[0077] An image acquisition module, which is used to acquire image information during the examination;

[0078] A three-dimensional coordinate module, which is communicatively connected to the image acquisition module, establishes a three-dimensional coordinate system based on the image information acquired by the image acquisition module, and calculates the three-dimensional coordinates of each point in the image;

[0079] An identification module, the identification module being communicatively connected to the three-dimensional coordinate module and used for identifying and calculating the motion trajectory of the examinee and the football according to the three-dimensional coordinates of each point in the image calculated by the three-dimensional coordinate module;

[0080] A judgment module, the judgment module being in communication connection with the identification module, and used to judge whether the examinee has committed a foul in the process of dribbling the ball around the pole and whether the ball has successfully scored a goal according to preset rules and the movement trajectory of the examinee and the ball identified by the identification module;

[0081] An analysis module, the analysis module being communicatively connected to the judgment module, and configured to automatically calculate the test scores according to the judgment results of the judgment module, and generate training suggestions for the examinees;

[0082] The data management module is connected to the analysis module in communication, stores the test data, and generates a test taker's score report and a test taker's training suggestion report.

[0083] Example: Figure 1-3 As shown, in this embodiment, a school is conducting a football test.

[0084] The camera angle adjustment unit allocates the shooting angle of each camera according to the layout of the examination venue and the position distribution of the cameras;

[0085] The camera parameter adjustment unit adjusts the camera, frame rate, resolution, exposure time, gain and white balance parameters. For locations in the center of the venue, the frame rate and resolution are increased, and for cameras in the edge areas of the venue, the frame rate and resolution are reduced; when the ambient light is strong, the exposure time is automatically shortened, and when the ambient light is weak, the gain is increased and the exposure time is extended.

[0086] The 3D coordinate module uses the image information collected by the image acquisition module to calculate the 3D coordinates of each point in the image by matching the parallax of corresponding points in different camera images;

[0087] First, establish an accurate three-dimensional coordinate system for the test site, set up the x-axis, y-axis and z-axis; confirm the coordinate points in the three-dimensional coordinate system through any two cameras in the camera module;

[0088] The optical center of the first camera is O1, and the intersection of the camera optical axis and the imaging plane, i.e. the principal point coordinates, is (x o1 ,y o1 ), the focal length is f1; the optical center of the second camera is O2, the intersection of the camera optical axis and the imaging plane, the principal point coordinates are (x o2 ,y o2 ), focal length is f2;

[0089] b is the distance between the optical centers O1 and O2 of the two cameras, i.e., the baseline distance b;

[0090] For a point P in space, the horizontal and vertical coordinates of the imaging plane of the first camera are (x1, y1), and the coordinates of the imaging plane of the second camera are (x2, y2). The parallax is the difference between the horizontal coordinates of the same spatial point in the imaging planes of the two cameras, d = (x1-x2);

[0091] The coordinates of the spatial point P in the first camera coordinate system are (x p1 ,y p1 ,z p1 );

[0092] According to the principle of similar triangles and triangulation, we have the following formula:

[0093]

[0094] From formula (1), we can get:

[0095]

[0096] Substituting (3) into (2), we get:

[0097]

[0098] Substituting equation (5) into equation (3) and equation (4) respectively, we can obtain x p1 andp1 According to this method, the coordinates of each point photographed by the camera in the three-dimensional coordinate system can be established.

[0099] Through the camera angle adjustment unit and camera parameter adjustment unit in the image acquisition module, the shooting angle can be flexibly allocated and the camera parameters can be intelligently adjusted according to the layout of the examination venue and the camera position distribution to ensure the collection of high-quality image information. The three-dimensional coordinate module calculates the three-dimensional coordinates of each point in the image to achieve accurate spatial modeling and data collection of the examination scene, greatly improving the spatial perception ability and data accuracy of the examination scene.

[0100] After establishing the coordinates of each point photographed by the camera in the three-dimensional coordinate system, the coordinate extraction unit is used to extract the coordinate points of the basketball and the human body;

[0101] The human motion recognition unit calculates the human motion trajectory according to the coordinates of the key points of the human skeleton extracted by the coordinate extraction unit;

[0102] The human body is based on skeleton recognition technology to extract skeleton key points, and obtain the coordinates of these key points in the three-dimensional coordinate system to calculate the human body movement trajectory.

[0103] The human action recognition unit is based on the specified time points t1, t2, t3, ..., t n And the corresponding three-dimensional coordinates of the skeleton key points Calculate the displacement between adjacent time points in sequence:

[0104] For the time period [t1, t2], the displacement Δs 1,2 for:

[0105]

[0106] Where s represents the displacement, That is, it is expressed as a displacement change vector;

[0107] The direction vector is:

[0108]

[0109] Repeating formula (8) at each different adjacent time point can obtain the displacement between each two adjacent time points: and direction vector Δs 2,3 ; Where i and j represent two adjacent time points; represents the direction vector;

[0110] Starting point As the starting point, each point is connected in sequence according to the calculated direction vector and displacement to obtain the motion trajectory of each key point of the human body within the specified time, thereby obtaining the motion trajectory of the human body.

[0111] The football trajectory recognition unit determines the position change trajectory of the football by matching the similarity of the football boundary coordinate vectors in adjacent frame images and then calculating the position change vector.

[0112] The football boundary coordinates are extracted based on shape detection and color detection technology, and the coordinates of these football boundary points in the three-dimensional coordinate system are obtained to calculate the football's motion trajectory. At the kth frame, the three-dimensional coordinates of several points on the football boundary form a boundary coordinate vector

[0113] The football trajectory recognition unit calculates the similarity of the football boundary coordinate vectors in two adjacent frames of images;

[0114]

[0115] in, is the inner product of two eigenvectors, are the moduli of the two feature vectors respectively; if the similarity of the feature vectors of two adjacent frames of football is greater than 0.9, they are considered to be matched;

[0116] After determining that the football in the adjacent frame images is a matching object, calculate the position change of the football:

[0117] The coordinates of the football in the kth frame image are (x k ,y k ,z k ), the coordinates of the football in the k+1th frame image are (x k91 ,y k91 ,z k91 ); the position change vector is:

[0118]

[0119] The matching calculation formula (9) and the position change formula (10) are calculated for all adjacent frames to obtain the position change trajectory of the football in the entire image sequence.

[0120] Through the coordinated work of the coordinate extraction unit, human motion recognition unit and football trajectory recognition unit in the recognition module, the coordinates of the key points of the human skeleton and the boundary coordinates of the football are accurately extracted, and the human motion trajectory and the football motion trajectory are accurately calculated, providing a more accurate and detailed basis, which helps to improve the fairness and objectivity of the test evaluation.

[0121] The judgment module uses the following rules to judge violations, including but not limited to:

[0122] Judgment of a foul on the pole: When the distance between the human body and the pole is less than or equal to 0, or the trajectory of the ball deviates from the prescribed route around the pole, it is judged as a foul on the pole; when it is greater than 0 or the trajectory of the ball does not deviate from the route around the pole, it is judged as a foul on not going around the pole;

[0123] Handball foul judgment: When the coordinates of the key points of the hand skeleton are detected to coincide with the coordinates of the football boundary or the distance is less than zero, it is judged as a handball foul; when they do not coincide or the distance is greater than zero, it is judged as no handball foul;

[0124] Goal judgment: Through straight line detection, the position and direction of the goal line are determined. When the boundary coordinates of the football cross the goal line, it is judged as a goal. If it does not cross, it is judged as a goal not scored.

[0125] The analysis module generates performance analysis and training suggestions;

[0126] Performance analysis includes but is not limited to foul penalty calculation and goal bonus calculation;

[0127] Calculation of foul deduction points: Based on the judgment result of the judgment module, for each foul behavior, points are deducted according to the preset deduction rules;

[0128] Calculation of goal bonus: If the candidate successfully shoots the football into the goal, he / she will be awarded bonus points according to the difficulty, distance and angle of the goal;

[0129] The training suggestions generate targeted training suggestions based on the judgment results of the judgment module.

[0130] The storage unit stores the results calculated by each module of the system separately; the display unit is used to display the candidate score report and candidate training suggestion report generated according to the analysis results of the analysis module.

[0131] Through the cooperation of the judgment module, analysis module and data management module, the judgment module judges foul behaviors and goal situations according to preset rules, the analysis module automatically calculates the test scores and generates training suggestions based on this, and the data management module stores the relevant data and generates reports, making the entire test process more scientific and standardized. Candidates can clearly understand their test performance and shortcomings, providing a clear direction for subsequent training, and it is also convenient for the test management department to uniformly manage and analyze the test data.

[0132] The embodiments of the present invention are provided for the purpose of illustration and description. Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations of the present invention. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A football examination system based on AI vision, characterized in that: include: An image acquisition module, which is used to acquire image information during the examination; A three-dimensional coordinate module, which is communicatively connected to the image acquisition module, establishes a three-dimensional coordinate system based on the image information acquired by the image acquisition module, and calculates the three-dimensional coordinates of each point in the image; An identification module, the identification module being communicatively connected to the three-dimensional coordinate module and used for identifying and calculating the motion trajectory of the examinee and the football according to the three-dimensional coordinates of each point in the image calculated by the three-dimensional coordinate module; A judgment module, the judgment module being in communication connection with the identification module, and used to judge whether the examinee has committed a foul in the process of dribbling the ball around the pole and whether the ball has successfully scored a goal according to preset rules and the movement trajectory of the examinee and the ball identified by the identification module; An analysis module, the analysis module being communicatively connected to the judgment module, and configured to automatically calculate the test scores according to the judgment results of the judgment module, and generate training suggestions for the examinees; The data management module is connected to the analysis module in communication, stores the test data, and generates a test taker's score report and a test taker's training suggestion report.

2. A football examination system based on AI vision as claimed in claim 1, characterized in that: The image acquisition module includes a camera angle adjustment unit and a camera parameter adjustment unit; The camera angle adjustment unit is used to allocate the shooting angle of each camera according to the layout of the examination venue and the position distribution of the cameras; The camera parameter adjustment unit is used to adjust the camera parameters including but not limited to frame rate, resolution, exposure time, gain and white balance. For locations in the center of the venue, the frame rate and resolution are increased, and for cameras in the edge area of ​​the venue, the frame rate and resolution are reduced; when the ambient light is strong, the exposure time is automatically shortened, and when the ambient light is weak, the gain is increased and the exposure time is extended.

3. A football examination system based on AI vision as claimed in claim 1, characterized in that: The three-dimensional coordinate module uses the image information collected by the image acquisition module to calculate the three-dimensional coordinates of each point in the image by matching the parallax of corresponding points in different camera images; First, establish an accurate three-dimensional coordinate system for the test site, set up the x-axis, y-axis and z-axis; confirm the coordinate points in the three-dimensional coordinate system through any two cameras in the camera module; The optical center of the first camera is O1, and the intersection of the camera optical axis and the imaging plane, i.e. the principal point coordinates, is (x o1 ,y o1 ), the focal length is f1; the optical center of the second camera is O2, the intersection of the camera optical axis and the imaging plane, the principal point coordinates are (x o2 ,y o2 ), focal length is f2; b is the distance between the optical centers O1 and O2 of the two cameras, i.e., the baseline distance b; For a point P in space, the horizontal and vertical coordinates of the imaging plane of the first camera are (x1, y1), and the coordinates of the imaging plane of the second camera are (x2, y2). The parallax is the difference between the horizontal coordinates of the same spatial point in the imaging planes of the two cameras, d = (x1-x2); The coordinates of the spatial point P in the first camera coordinate system are (x p1 ,y p1 ,z p1 ); According to the principle of similar triangles and triangulation, we have the following formula: From formula (1), we can get: Substituting (3) into (2), we get: Substituting equation (5) into equation (3) and equation (4) respectively, we can obtain x p1 and p1 According to this method, the coordinates of each point photographed by the camera in the three-dimensional coordinate system can be established.

4. A football examination system based on AI vision as claimed in claim 1, characterized in that: The recognition module includes a coordinate extraction unit, a human motion recognition unit and a football trajectory recognition unit; The coordinate extraction unit is used to extract coordinate points of the basketball and the human body; The human motion recognition unit calculates the human motion trajectory according to the coordinates of the human skeleton key points extracted by the coordinate extraction unit; The football trajectory recognition unit determines the position change trajectory of the football by matching the similarity of the football boundary coordinate vectors in adjacent frame images and then calculating the position change vector.

5. A football examination system based on AI vision as claimed in claim 4, characterized in that: The coordinate extraction unit extracts the coordinates of the football and the human body; The football boundary coordinates are extracted based on shape detection and color detection technology, and the coordinates of these football boundary points in the three-dimensional coordinate system are obtained to calculate the football's motion trajectory. At the kth frame, the three-dimensional coordinates of several points on the football boundary form a boundary coordinate vector The human body is based on skeleton recognition technology to extract skeleton key points, and obtain the coordinates of these key points in the three-dimensional coordinate system to calculate the human body movement trajectory.

6. A football examination system based on AI vision as claimed in claim 5, characterized in that: The human action recognition unit is based on the specified time points t1, t2, t3, ..., t n And the corresponding three-dimensional coordinates of the skeleton key points Calculate the displacement between adjacent time points in sequence: For the time period [t1, t2], the displacement Δs 1,2 for: Where s represents the displacement, That is, it is expressed as a displacement change vector; The direction vector is: Repeating formula (8) at each different adjacent time point can obtain the displacement between each two adjacent time points: and direction vector ΔS 2,3 ; Where i and j represent two adjacent time points; represents the direction vector; Starting point As the starting point, each point is connected in sequence according to the calculated direction vector and displacement to obtain the motion trajectory of each key point of the human body within the specified time, thereby obtaining the motion trajectory of the human body.

7. A football examination system based on AI vision as claimed in claim 5, characterized in that: The football trajectory recognition unit calculates the similarity of the football boundary coordinate vectors in two adjacent frames of images; in, is the inner product of two eigenvectors, are the moduli of the two feature vectors respectively; if the similarity of the feature vectors of two adjacent frames of football is greater than 0.9, they are considered to be matched; After determining that the football in the adjacent frame images is a matching object, calculate the position change of the football: The coordinates of the football in the kth frame image are (x k ,y k ,z k ), the coordinates of the football in the k+1th frame image are (x k91 ,y k91 ,z k91 ); the position change vector is: The matching calculation formula (9) and the position change formula (10) are calculated for all adjacent frames to obtain the position change trajectory of the football in the entire image sequence.

8. The football examination system based on AI vision as claimed in claim 1, characterized in that: The judgment module determines violations by using, but is not limited to, the following rules: Judgment of a foul on the pole: When the distance between the human body and the pole is less than or equal to 0, or the trajectory of the ball deviates from the prescribed route around the pole, it is judged as a foul on the pole; when it is greater than 0 or the trajectory of the ball does not deviate from the route around the pole, it is judged as a foul on not going around the pole; Handball foul judgment: When the coordinates of the key points of the hand skeleton are detected to coincide with the coordinates of the football boundary or the distance is less than zero, it is judged as a handball foul; when they do not coincide or the distance is greater than zero, it is judged as no handball foul; Goal judgment: Through straight line detection, the position and direction of the goal line are determined. When the boundary coordinates of the football cross the goal line, it is judged as a goal. If it does not cross, it is judged as a goal not scored.

9. The football examination system based on AI vision as claimed in claim 1, characterized in that: The analysis module generates performance analysis and training suggestions; The performance analysis includes but is not limited to the calculation of foul deduction points and goal bonus points; The foul deduction calculation: according to the judgment result of the judgment module, for each foul behavior, points are deducted according to the preset deduction rules; Calculation of goal bonus: If the candidate successfully shoots the football into the goal, corresponding bonus points will be awarded according to the difficulty, distance and angle of the goal; The training suggestion generates targeted training suggestions according to the determination result of the determination module.

10. The football examination system based on AI vision as claimed in claim 1, characterized in that: The data management module includes a storage unit and a display unit; The storage unit stores the results calculated by each module of the system respectively; The display unit is used to display the examinee's score report and examinee's training suggestion report generated according to the analysis result of the analysis module.

Images