Table tennis teaching action judgment and prompting method and system

Abstract

The present invention relates to the field of sports teaching technology, and in particular to a table tennis teaching action judgment and prompting method and system, the method comprising: classifying M individual techniques based on structural similarity to obtain N action patterns; each action pattern corresponds to P basic questions; each basic question corresponds to a calculation function; obtaining a time series video during a trainee's training process; obtaining an action pattern corresponding to the time series video based on the time series video and a judgment function; extracting from the time series video based on P calculation functions corresponding to the action pattern corresponding to the time series video to obtain demand data; calculating based on the demand data and the P calculation functions to obtain P calculation values; comparing each calculation value with a corresponding set threshold to obtain whether the action pattern corresponding to the time series video is qualified. The present invention improves the accuracy and response speed of action judgment and has better practicality.

Description

Technical Field

[0001] The invention relates to the technical field of sports teaching, and in particular to a table tennis teaching action judgment and prompting method and system. Background Art

[0002] At present, there are more and more methods for computer vision to assist physical education teaching, most of which are to compare user videos with standard action libraries and obtain scores by calculating similarities. This method is transferred from the field of fitness. When used for ball sports, it ignores the movement rules of different balls. It is necessary to establish a standard action library in advance and do manual labeling. When comparing, it is necessary to use neural networks for image classification first, and it is also necessary to stretch and transform the human body model to eliminate the impact of the difference between the human body size and the model in the standard action library. It is also necessary to establish multiple multi-layer convolutional networks to reduce the deviation caused by the limitations of the model sample. It consumes a lot of software and hardware computing resources, has high requirements for the accuracy of data acquisition, and has limited accuracy, slow response speed, and low practicality.

[0003] Especially for table tennis, there is no standard action. Players with the same technique but different playing styles have very different actions. For example, for the same forehand attack technique, the forehand attack action of a slice player and an attack player is different.

[0004] Even players with the same playing style can have very different movements due to different innate conditions and training experiences. If one person's movements are used as the standard movements to judge the movements of others, the movements of others will be judged as non-standard.

[0005] Even for the same person, when judging whether his action is reasonable, it is also necessary to consider it in conjunction with the incoming ball; for example, if the incoming ball is fast but the racket is slow, even if the racket-pulling action itself is "standard", the overall hitting process is unreasonable and it is very likely to result in a mistake.

[0006] Even for the same ball, the same person may make different technical adjustments, such as more friction or more impact, which may be reasonable actions, but should not be evaluated with standard actions. Summary of the invention

[0007] In view of this, the present invention provides a table tennis teaching action judgment and prompting method and system, which focuses on reasonable actions rather than standards, is more in line with the movement laws of table tennis, is more scientific, and greatly saves computing resources, improves accuracy, response speed and practicality, and is also more convenient for deployment on mobile terminals.

[0008] To achieve the above object, the present invention provides the following solutions:

[0009] A table tennis teaching action judgment and prompting method, comprising:

[0010] Based on the structural similarity, M individual technologies are classified to obtain N action modes; each action mode corresponds to P basic problems; each basic problem corresponds to a calculation function; M, N and P are all positive integers greater than or equal to 1;

[0011] Obtaining a time series video of a trainee during training; obtaining an action mode corresponding to the time series video based on the time series video and a judgment function;

[0012] Extracting from the time series video based on P calculation functions corresponding to the action mode corresponding to the time series video to obtain demand data;

[0013] Perform calculation based on the demand data and the P calculation functions to obtain P calculation values;

[0014] Each of the calculated values ​​is compared with the corresponding set threshold value to determine whether the action mode corresponding to the time series video is qualified; if unqualified, the unqualified basic problem is output.

[0015] Preferably, each of the basic problems corresponds to Q auxiliary functions;

[0016] When the calculated value cannot be obtained by calculation based on the demand data and the calculation function, auxiliary data is extracted from the time series video based on Q auxiliary functions; and the calculated value is obtained by calculation based on the auxiliary data and Q auxiliary functions.

[0017] Preferably, the N action modes are respectively forehand topspin out of the table, forehand backspin out of the table, forehand inside topspin out of the table, forehand inside backspin out of the table, backhand topspin out of the table, backhand backspin out of the table, backhand inside topspin out of the table and backhand inside backspin out of the table.

[0018] Preferably, the judgment function includes a main function and O candidate functions;

[0019] When the action mode corresponding to the time series video can be obtained based on the time series video and the main function, the action mode corresponding to the time series video is obtained based on the time series video and the main function; if not, the action mode corresponding to the time series video is obtained based on O alternative functions and the time series video.

[0020] The present invention also provides a table tennis teaching action judgment and prompting system, comprising:

[0021] A classification module, used for classifying M individual technologies based on structural similarity to obtain N action modes; each of the action modes corresponds to P basic problems; each of the basic problems corresponds to a calculation function; M, N and P are all positive integers greater than or equal to 1;

[0022] A data acquisition module is used to acquire the time series video of the trainee during training; based on the time series video and the judgment function, the action mode corresponding to the time series video is obtained;

[0023] A data extraction module, used to extract required data from the time series video based on P calculation functions corresponding to the action mode corresponding to the time series video;

[0024] A calculation module, used for performing calculation based on the demand data and the P calculation functions to obtain P calculation values;

[0025] The judgment module is used to compare each of the calculated values ​​with the corresponding set threshold value to determine whether the action mode corresponding to the time series video is qualified; if unqualified, the unqualified basic problem is output.

[0026] Preferably, each of the basic problems corresponds to Q auxiliary functions;

[0027] When the calculation module is unable to obtain the calculated value based on the demand data and the calculation function, the data extraction module extracts from the time series video based on Q auxiliary functions to obtain auxiliary data; the calculation module performs calculation based on the auxiliary data and Q auxiliary functions to obtain the calculated value.

[0028] Preferably, the N action modes are respectively forehand topspin out of the table, forehand backspin out of the table, forehand inside topspin out of the table, forehand inside backspin out of the table, backhand topspin out of the table, backhand backspin out of the table, backhand inside topspin out of the table and backhand inside backspin out of the table.

[0029] Preferably, the judgment function includes a main function and O candidate functions;

[0030] When the data acquisition module can obtain the action mode corresponding to the time series video based on the time series video and the main function, the data acquisition module obtains the action mode corresponding to the time series video based on the time series video and the main function; if not, the data acquisition module obtains the action mode corresponding to the time series video based on O alternative functions and the time series video.

[0031] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0032] The present invention relates to a table tennis teaching action judgment and prompting method and system, the method comprising: classifying M single techniques based on structural similarity to obtain N action modes; each action mode corresponds to P basic problems; each basic problem corresponds to a calculation function; obtaining a time series video during the training process of a trainee; obtaining the action mode corresponding to the time series video based on the time series video and the judgment function; extracting from the time series video based on the P calculation functions corresponding to the action mode corresponding to the time series video to obtain demand data; calculating based on the demand data and the P calculation functions to obtain P calculation values; comparing each calculation value with a corresponding set threshold to obtain whether the action mode corresponding to the time series video is qualified. The present invention focuses on reasonable actions rather than standards, is more in line with the movement law of table tennis, is more scientific, and greatly saves computing resources, improves accuracy, response speed and practicality, and is more convenient for the deployment of mobile terminals. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0034] Figure 1 This is a flow chart of the table tennis teaching action judgment and prompting method of the present invention;

[0035] Figure 2 This is a structural diagram of the table tennis teaching action judgment and prompting system of the present invention;

[0036] Figure 3 The basic problem of the present invention is whether the center of gravity of the basic position is too high.

[0037] Figure 4 The basic problem of the present invention is whether the hitting point is too close to the body when calculating the data that the function needs to obtain;

[0038] Figure 5 The basic problem of the present invention is whether the racket shape of the ball is too normal to calculate the data required by the function;

[0039] Figure 6 The basic problem of the present invention is whether the center of gravity of the basic standing position is too high and the data that the auxiliary function needs to obtain is a schematic diagram.

[0040] Explanation of symbols: 1-classification module, 2-data acquisition module, 3-data extraction module, 4-calculation module, 5-judgment module. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] The purpose of the present invention is to provide a table tennis teaching action judgment and prompting method and system, which focuses on reasonable actions rather than standards, is more in line with the movement laws of table tennis, is more scientific, greatly saves computing resources, improves accuracy, response speed and practicality, and is more convenient for deployment on mobile terminals.

[0043] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0044] Figure 1 This is a flow chart of the table tennis teaching action judgment and prompting method of the present invention. Figure 1 As shown, the present invention provides a table tennis teaching action judgment and prompting method, comprising:

[0045] Step S1, classify M individual technologies based on structural similarity to obtain N action modes; each of the action modes corresponds to P basic problems; each of the basic problems corresponds to a calculation function; M, N and P are all positive integers greater than or equal to 1.

[0046] In this embodiment, the N action modes are respectively forehand topspin, forehand backspin, forehand inside topspin, forehand inside backspin, backhand topspin, backhand backspin, backhand inside topspin and backhand inside backspin. Specifically, the action mode may also include serving actions, such as whether the served ball has topspin / backspin and whether the served ball has left spin / right spin.

[0047] For example, forehand topspin includes forehand attack, forehand topspin, forehand backspin, forehand quick drive, forehand backspin, forehand high ball, forehand defense and forehand half-out of the court.

[0048] Step S2, obtaining a time series video of the trainee during training; and obtaining an action pattern corresponding to the time series video based on the time series video and a judgment function.

[0049] Wherein, the judgment function includes a main function and O candidate functions.

[0050] When the action mode corresponding to the time series video can be obtained based on the time series video and the main function, the action mode corresponding to the time series video is obtained based on the time series video and the main function; if not, the action mode corresponding to the time series video is obtained based on O alternative functions and the time series video.

[0051] Select any one of the O candidate functions to obtain the action mode corresponding to the time series video.

[0052] Preferably, in order to save computing resources and speed up the response, when obtaining the action mode corresponding to the time series video, any frame of image in the time series video is selected for calculation with the judgment function.

[0053] For example: suppose the length of the table is L. At the moment of touching the ball, the X-axis coordinate values ​​of the wrist, elbow and shoulder joints of the student's racket-holding hand are Xa, Xb and Xc respectively, the Y-axis coordinate values ​​of the highest point H on the upper edge of the racket and the lowest point L on the lower edge are Yh and Yl respectively, and the Y-axis coordinate value of the incoming ball position is Yp. If an image frame satisfies: Xa>Xb>Xc (indicating that the shoulder, elbow and wrist are to the right in turn), Yl>Yh (indicating that the racket face is tilted forward) and Yp>L / 2 (indicating that the hitting point is outside the table), then the image frame is classified as the action mode of "forehand topspin off the table".

[0054] Step S3, extracting from the time series video based on the P calculation functions corresponding to the action mode corresponding to the time series video to obtain the required data.

[0055] For example, when the basic question is whether the center of gravity of the basic position is too high, the calculation function is (E+ho) / F, where E is the height of the center of mass of the position, F is the height, and ho is the height of the origin of the coordinate system relative to the ground; then the height and height of the center of mass of the position are extracted from a certain image frame in the time series video, such as Figure 3 shown.

[0056] Preferably, a set number of image frames can be optionally extracted from the time series video and then the average value can be calculated to obtain the height of the centroid and the body height of the standing position.

[0057] For another example, when the basic question is whether the hitting point is too close to the body, the calculation function is the vector product to calculate the angle θ between the position point of the incoming ball, the left shoulder joint point and the right shoulder joint point at the moment of contact (i.e., the point of contact). Then, the coordinate values ​​of the position point of the incoming ball, the left shoulder joint point and the right shoulder joint point at the moment of contact are extracted from the time series video, such as Figure 4 Whether to select from a single frame image or multiple frames of images depends on actual needs.

[0058] For another example, when the action mode is forehand backspin, the basic question is whether the racket shape of the rub is too flat, and the calculation function is acos((Yh-Yl) / sqrt(2*(Zh-Zl)+2*(Yh-Yl))), where Yh is the Y-axis coordinate value of the highest point H on the upper edge of the racket, Zh is the Z-axis coordinate value of the highest point H on the upper edge of the racket, Yl is the Y-axis coordinate value of the lowest point L on the lower edge of the racket, Zl is the Z-axis coordinate value of the lowest point L on the lower edge of the racket, acos is the arccosine function, and sqrt is the square root calculation function. The required data is extracted from the time series video, such as Figure 5 As shown, whether to select from a single frame image or multiple frame images is based on actual needs.

[0059] Step S4, performing calculation based on the demand data and the P calculation functions to obtain P calculation values.

[0060] Preferably, each of the basic problems corresponds to Q auxiliary functions; when the calculated value cannot be obtained by calculation based on the demand data and the calculation function, auxiliary data is extracted from the time series video based on the Q auxiliary functions to obtain the auxiliary data; and the calculated value is obtained by calculation based on the auxiliary data and the Q auxiliary functions.

[0061] The calculated value can be obtained by performing calculation on any one of the Q auxiliary functions.

[0062] At the same time, the calculation function and the auxiliary function can also be used simultaneously to check each other and improve accuracy.

[0063] By setting calculation functions and auxiliary functions, the normal operation of the method can be guaranteed when data is missing (such as a key point is blocked), thereby reducing the requirements for hardware, shooting environment and computing load. For example, when the basic question is whether the center of gravity of the basic position is too high, the height of the center of mass of the position is not easy to obtain accurately due to limited conditions. The height of the center point of the line connecting the two shoulders and the height can also be obtained. The auxiliary function is ((Zleft+Zright) / 2+ho) / F, where Zleft is the height of the left shoulder and Zright is the height of the right shoulder. Figure 6 shown.

[0064] Step S5, comparing each of the calculated values ​​with the corresponding set threshold value to determine whether the action mode corresponding to the time series video is qualified; if unqualified, outputting the unqualified basic question. Step S5 is implemented based on a judgment function.

[0065] For example, when the basic question is whether the center of gravity of the basic position is too high, if (E+ho) / F> the set threshold, it can be judged that the center of gravity is too high.

[0066] Or, when the basic question is whether the hitting point is too close to the body, if the angle is > 60°, it is judged to be too close.

[0067] Alternatively, when the basic question is whether the racket shape is too flat when rubbing the ball, if acos((Yh-Yl) / sqrt(2*(Zh-Zl)+2*(Yh-Yl)))>the set threshold, that is, the racket face tilting angle is greater than the set threshold, it is judged that the racket shape is too flat.

[0068] The set threshold is not completely fixed, but can be adjusted dynamically as needed. Using different individual techniques in the same action mode, or facing different incoming balls with the same individual technique, as well as the evolution of technical trends and the students' own practice purposes can all trigger threshold adjustments, so as to grasp the laws of motion more scientifically, assist teaching judgments, and adapt to the students' personalized needs.

[0069] For different single technologies in the same action mode, the present invention does not pre-divide the video into different single technologies, because the basic problems and judgment functions of various single technologies in the same action mode are common, and there is no need to further subdivide and increase the amount of calculation. However, the threshold setting for the basic problem sometimes needs to be dynamically adjusted:

[0070] When the coach or student gives a single skill for this training, the corresponding threshold setting will be fine-tuned under the same action mode and the same judgment function for the same basic problem. For example, for the same action mode of "forehand topspin", to judge whether the student's action has the basic problem of "high center of gravity when running and hitting the ball", the threshold setting of "forehand hitting half-high ball" needs to be slightly higher than that of "forehand topspin".

[0071] When there is no single technique for this training, the corresponding threshold will be automatically fine-tuned according to the incoming ball data. For example, to determine whether the student's action has a basic problem such as "the center of gravity is too high when running and hitting the ball", the threshold will be automatically adjusted according to the height of the incoming ball arc. The higher the incoming ball, the higher the threshold.

[0072] Preferably, the deviation percentage of the value calculated by the judgment function relative to the preset absolute number threshold can be used to replace the preset absolute number threshold as the judgment criterion.

[0073] As technology trends evolve and students’ own practice objectives change, the thresholds for basic questions sometimes need to be adjusted dynamically:

[0074] Table tennis is developing towards a stronger confrontation intensity, relatively weaker rotation, and more balanced forehand and backhand, so the threshold setting needs to be adjusted accordingly. For example, in the past, forehand was the main and backhand was the auxiliary, and the basic stance was left foot in front and right foot in back, but now the emphasis is on balance on both sides, and the basic stance is more inclined to have two feet parallel. The function for judging the front and back position difference of the left and right feet in the basic stance, the threshold setting now needs to be closer to 0.

[0075] Whether to adjust the threshold setting also depends on the individual situation of the students. For example, if we still use the function of judging the front and rear position difference of the left and right feet in the basic stance, for young students, the threshold setting needs to be closer to 0 now, because their learning is for the future; while for adult amateurs, the threshold setting can remain unchanged (obviously greater than 0), because their purpose is entertainment and fitness. If the basic stance is forcibly adjusted to imitate the latest trend, it may affect the whole body and cause the entire movement system to deform.

[0076] Furthermore, the unqualified basic questions are output in multimedia forms including voice, text, pictures, animations, videos, VR, AR, MR and XR.

[0077] Preferably, the scores of various action modes can be given by action mode, the scores can be given by action link, or the weighted average total score can be given. The score can be based on a set threshold, or it can be based on its own historical training data, or it can be based on the average level of multiple students. The score can be a certain value, or it can be replaced by expressions and pictures. The score can be based only on the student's action data as the main data source, or it can be combined with the trajectory data of the ball after the student actually hits the ball.

[0078] Figure 2 This is a structural diagram of the table tennis teaching action judgment and prompting system of the present invention. Figure 2 As shown, the present invention provides a table tennis teaching action judgment and prompting system, including: a classification module 1, a data acquisition module 2, a data extraction module 3, a calculation module 4 and a judgment module 5.

[0079] The classification module 1 is used to classify M individual technologies based on structural similarity to obtain N action modes; each of the action modes corresponds to P basic problems; each of the basic problems corresponds to a calculation function; M, N and P are all positive integers greater than or equal to 1.

[0080] The data acquisition module 2 is used to acquire the time series video of the trainees during training; and obtain the action mode corresponding to the time series video based on the time series video and the judgment function.

[0081] The data extraction module 3 is used to extract the required data from the time series video based on the P calculation functions corresponding to the action mode corresponding to the time series video.

[0082] The calculation module 4 is used to perform calculations based on the demand data and the P calculation functions to obtain P calculation values.

[0083] The judgment module 5 is used to compare each of the calculated values ​​with the corresponding set threshold value to determine whether the action mode corresponding to the time series video is qualified; if unqualified, the unqualified basic problem is output.

[0084] Preferably, each of the basic problems corresponds to Q auxiliary functions.

[0085] When the calculation module 4 is unable to obtain the calculated value based on the demand data and the calculation function, the data extraction module 3 extracts from the time series video based on Q auxiliary functions to obtain auxiliary data; the calculation module 4 performs calculation based on the auxiliary data and Q auxiliary functions to obtain the calculated value.

[0086] Preferably, the N action modes are respectively forehand topspin out of the table, forehand backspin out of the table, forehand inside topspin out of the table, forehand inside backspin out of the table, backhand topspin out of the table, backhand backspin out of the table, backhand inside topspin out of the table and backhand inside backspin out of the table.

[0087] Preferably, the judgment function includes a main function and O candidate functions.

[0088] When the data acquisition module 2 can obtain the action mode corresponding to the time series video based on the time series video and the main function, the data acquisition module 2 obtains the action mode corresponding to the time series video based on the time series video and the main function; if not, the data acquisition module 2 obtains the action mode corresponding to the time series video based on O alternative functions and the time series video.

[0089] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0090] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A table tennis teaching action judgment and prompting method, characterized in that: include: Based on the structural similarity, M individual technologies are classified to obtain N action modes; each of the action modes corresponds to P basic problems; each of the basic problems corresponds to a calculation function; M, N and P are all positive integers greater than or equal to 1; Obtain time series video of trainees during training; Obtaining an action mode corresponding to the time series video based on the time series video and a judgment function; Extracting from the time series video based on P calculation functions corresponding to the action mode corresponding to the time series video to obtain demand data; Calculation is performed based on the required data and the P calculation functions to obtain P calculation values; wherein each of the basic problems corresponds to Q auxiliary functions; when the calculation value cannot be calculated based on the required data and the calculation function, auxiliary data is extracted from the time series video based on the Q auxiliary functions; calculation is performed based on the auxiliary data and the Q auxiliary functions to obtain the calculation value; Each of the calculated values ​​is compared with the corresponding set threshold value to determine whether the action mode corresponding to the time series video is qualified; if unqualified, the unqualified basic problem is output.

2. The table tennis teaching action judgment and prompting method according to claim 1, characterized in that: The N action modes are respectively forehand out-of-table topspin, forehand out-of-table backspin, forehand inside-table topspin, forehand inside-table backspin, backhand out-of-table topspin, backhand out-of-table backspin, backhand inside-table topspin and backhand inside-table backspin.

3. The table tennis teaching action judgment and prompting method according to claim 1, characterized in that: The judgment function includes a main function and O candidate functions; When the action mode corresponding to the time series video can be obtained based on the time series video and the main function, the action mode corresponding to the time series video is obtained based on the time series video and the main function; if not, the action mode corresponding to the time series video is obtained based on O alternative functions and the time series video.

4. A table tennis teaching action judgment and prompting system, characterized in that: include: A classification module, used to classify M individual technologies based on structural similarity to obtain N action modes; each of the action modes corresponds to P basic problems; each of the basic problems corresponds to a calculation function; M, N and P are all positive integers greater than or equal to 1; The data acquisition module is used to obtain the time series video of the trainees during their training; Obtaining an action mode corresponding to the time series video based on the time series video and a judgment function; A data extraction module, used to extract required data from the time series video based on P calculation functions corresponding to the action mode corresponding to the time series video; A calculation module, configured to perform calculations based on the required data and the P calculation functions to obtain P calculation values; wherein each of the basic problems corresponds to Q auxiliary functions; when the calculation module cannot perform calculations based on the required data and the calculation functions to obtain the calculation values, the data extraction module extracts from the time series video based on the Q auxiliary functions to obtain auxiliary data; the calculation module performs calculations based on the auxiliary data and the Q auxiliary functions to obtain the calculation values; The judgment module is used to compare each of the calculated values ​​with the corresponding set threshold value to determine whether the action mode corresponding to the time series video is qualified; if unqualified, the unqualified basic problem is output.

5. The table tennis teaching action judgment and prompting system according to claim 4, characterized in that: The judgment function includes a main function and O candidate functions; When the data acquisition module can obtain the action mode corresponding to the time series video based on the time series video and the main function, the data acquisition module obtains the action mode corresponding to the time series video based on the time series video and the main function; if not, the data acquisition module obtains the action mode corresponding to the time series video based on O alternative functions and the time series video.

Images