CN110929595A - System and method for training or entertainment with or without ball based on artificial intelligence - Google Patents
System and method for training or entertainment with or without ball based on artificial intelligence Download PDFInfo
- Publication number
- CN110929595A CN110929595A CN201911080440.1A CN201911080440A CN110929595A CN 110929595 A CN110929595 A CN 110929595A CN 201911080440 A CN201911080440 A CN 201911080440A CN 110929595 A CN110929595 A CN 110929595A
- Authority
- CN
- China
- Prior art keywords
- player
- ball
- detection
- training
- action
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 64
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 21
- 238000001514 detection method Methods 0.000 claims abstract description 139
- 230000009471 action Effects 0.000 claims abstract description 94
- 230000000007 visual effect Effects 0.000 claims abstract description 23
- 230000033001 locomotion Effects 0.000 claims abstract description 19
- 238000013135 deep learning Methods 0.000 claims description 15
- 230000000694 effects Effects 0.000 claims description 11
- 238000013527 convolutional neural network Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 4
- 230000003287 optical effect Effects 0.000 claims description 4
- 238000002372 labelling Methods 0.000 claims description 3
- 210000003414 extremity Anatomy 0.000 description 9
- 230000036541 health Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 210000000245 forearm Anatomy 0.000 description 1
- 210000004247 hand Anatomy 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 210000000707 wrist Anatomy 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/23—Recognition of whole body movements, e.g. for sport training
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The invention discloses a system and a method for training or entertainment with or without a ball based on artificial intelligence, wherein the system comprises a player detection system, a ball detection system, a motion analysis system and a visual feedback system; when a player appears in the acquisition range, the player detection system learns player marking data in advance to obtain a player detection model, and starts a player detection method through the model to obtain detection frame information of the player under a pixel coordinate system; when balls appear in an acquisition range, a ball detection system learns ball marking data in advance to obtain a ball detection model, and obtains detection frames of players and positions of the balls in the pixel coordinate system; the motion analysis system predicts and calculates body joint points and limb positions of the players by combining information of the player detection boxes and the ball detection boxes; and the visual feedback system displays and feeds back the result of whether the player completes the preset action on the display equipment. The invention detects and identifies human bodies, balls and limbs of ball enthusiasts and guides and supervises the feedback of ball-in and ball-out action training.
Description
Technical Field
The invention relates to a ball and non-ball training system, in particular to a ball and non-ball training or entertainment system and method based on artificial intelligence.
Background
With the public attention to sports, the participation of basketball, football, volleyball and tennis is higher and higher, so the importance of ball-free physical ability and agility training for relevant skill and skillful action training, such as basketball dribbling and football pitching, is more and more obvious. In the training mode of the actions, one mode is an online teaching mode which is expensive and needs professional sites, and the other mode is that the actions of the public are simulated by watching videos and videos. Therefore, how to provide a training or entertainment system which can feed back and interact, is intelligent, convenient, easy to use and interesting for ball enthusiasts is a technical problem to be solved urgently.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to overcome the defects of the prior art and provides a system and a method for training or entertainment with or without balls based on artificial intelligence.
The technical scheme is as follows: the invention relates to a ball and non-ball training or entertainment system based on artificial intelligence, which comprises a player detection system, a ball detection system, a motion analysis system and a visual feedback system, wherein the player detection system comprises a player detection system, a ball detection system, a visual feedback system and a visual feedback system, wherein the player detection system comprises a player detection system, a ball detection system, a:
the player detection system learns player marking data in advance to obtain a player detection model when a player appears in the range of the view screen acquisition equipment, starts a player detection method through the model to obtain detection frame information of the player under a pixel coordinate system, and then tracks a player target by using a target tracking method, wherein the player detection method is a deep learning-based player detection method or a traditional detection method;
when balls appear in the acquisition range of acquisition equipment, a ball detection system learns ball labeling data in advance to obtain a ball detection model, starts a deep learning-based ball detection method or a traditional detection method through the model to obtain a ball detection frame and the position of the ball detection frame under a pixel coordinate system, and then tracks a ball target by using a target tracking method;
the motion analysis system is used for predicting and calculating body joint points and limb positions of the players according to the information of the player detection frames obtained by the player detection system and the information of the ball detection frames obtained by the ball detection system, and judging and analyzing the motions of the players;
and the visual feedback system displays and feeds back the result of whether the player finishes the preset action in the action analysis system on the display equipment.
The player detection method is used for training data labeled with a player target detection frame in a picture based on a convolutional neural network, and model parameters are optimized by using a back propagation method to obtain a player detection model.
When a target tracking method is used for tracking the player target, the convolutional neural network in deep learning is utilized to learn the target characteristics of the players in the labeled video data, and the characteristics of the players are extracted from the current frame and the previous frame and compared with the similarity.
When the player target is tracked by using a target tracking method, the position of a tracked player target frame in the next frame image is predicted based on a Fourier transform implementation method, a filtering method and a characteristic point-based optical flow method.
In the action analysis system, the prediction of the body joint points and the limb positions of the players trains the position information and the type information of the body joint points of the players through the method based on deep learning to obtain the position and the type information of different joint points of the bodies in a pixel coordinate system, the position information of the related limbs is obtained through the predicted position and type information of the joint points, and whether the players complete the preset actions is judged by combining a set action judgment method.
The expression formula of the action judging method is as follows:
n is the number of human body joint points and the position number of ball detection frames, i is the ith joint point or detection frame, Xi is the ith predicted joint point position information or detection frame position information, Xoi is the ith preset joint point position information or detection frame position informationτ is a predetermined threshold value.
The visual feedback system comprises a real-time player action feedback module, an end playback analysis module and an action scoring module; the real-time player action feedback module feeds back the result of each action of the player in real time in the training process and feeds back the direction of the correct action; the playback ending analysis module plays back the action after the training is ended; and the action scoring module is used for scoring the correct and wrong actions of the training.
A plurality of display modes are preset in the visual feedback system, and the display mode is a 'breakthrough' mode formed when the added action score reaches a preset threshold value and enters the next training; the feedback is displayed as a visual special effect and sound effect.
The artificial intelligence-based ball and non-ball training or entertainment method comprises the following steps:
(1) erecting a smart phone;
(2) adjusting the position of the player;
(3) the player makes a designated action;
(4) and (6) visualizing feedback.
Specifically, in the step (3), a basketball detection model in the ball detection system is called to obtain the position and size information of a basketball detection frame; and calling a player motion analysis system to obtain the information of body joint points and limb positions of the player, wherein the information is consistent with the information of a certain preset motion.
Has the advantages that: the invention has the following advantages:
(1) the system carries out human body detection and recognition, ball detection and recognition and limb detection and recognition on ball enthusiasts based on artificial intelligence so as to achieve the guidance and supervision feedback of ball and non-ball action training on the ball enthusiasts.
(2) The feedback of ball sports training is intelligentized by combining the most advanced artificial intelligence technology, the system can be applied to APP, and the acquisition and display equipment can be a smart phone, so that the system is more convenient to use.
(3) The action analysis system is used for evaluating the difference between the action of the player and the preset entertainment action or training action, accurately measuring the action completion degree and indicating the place with inaccurate action to be beneficial to the correction of the action of the player; for example, the related sports and actions of players with balls such as basketball volleyball, basketball screen contact, football juggling, ball-free agility training, physical training and the like can be collected by video collecting equipment, and the action accuracy can be judged by the system for training or entertainment and the like.
(4) The system increases the interest of training, and can preset a plurality of display modes in a terminal or other using equipment, and add action scores to reach a certain preset threshold value to enter the next more complex training to form a 'breakthrough' mode or add some visual special effects and sound effects and the like in the feedback display to greatly increase the use interest of a user.
Drawings
FIG. 1 is a system configuration diagram of the present invention;
FIG. 2 is a flow chart of an embodiment of the system of the present invention;
FIG. 3 is a diagram illustrating a smartphone placement location in accordance with an embodiment of the present invention;
FIG. 4 is a schematic diagram of player positions in an embodiment of the system of the present invention;
figure 5 is a schematic diagram of basketball touch training in an embodiment of the system of the present invention.
Detailed Description
As shown in figure 1, the present invention is based on an artificial intelligence ball and non-ball training or entertainment system, which comprises a player detection system, a ball detection system, a motion analysis system and a visual feedback system.
The player detection system starts a preset player detection method when a player appears in the acquisition range of the video acquisition equipment so as to train a player detection model in advance, and the detection method is a deep learning-based player detection method or a traditional detection method.
The player detection method based on deep learning is to learn basketball marking data by using a general target detection technology to obtain a basketball detection model for detecting players; specifically, training is carried out on data of a player target detection frame marked in a picture based on a convolutional neural network, and model parameters are optimized by using a back propagation algorithm to obtain a player detection model.
The traditional detection method is to use Harr and HOG characteristics, combine machine learning and adopt SVM and Adaboost methods to detect players.
In addition, a player tracking model needs to be prepared in advance to track the position of the player all the time, and the following two methods are mainly adopted:
1. learning the characteristics of the player target in the marked video data by using a convolutional neural network in deep learning, extracting the characteristics of the player from the current frame and the previous frame when in use, and comparing the similarity to predict the position change of the target in the next frame;
2. the method based on Fourier transform implementation, the filtering method and the optical flow method based on the feature points predict and track the position of the target frame of the player in the next frame image.
When the detection device is used, when a player appears in the range of the screen acquisition equipment, a pre-trained player detection method is started to detect a player target frame, then a target tracking method is used for tracking a player target, and the position and size information of the player in the detection frame of the current frame is obtained all the time.
A ball detection system trains a player detection model by using a ball detection method when balls appear in the acquisition range of acquisition equipment, and the method comprises a deep learning-based method or a traditional detection method.
Specifically, a ball detection model is trained in advance, and the training method is divided into two types:
1. training the data marked with the ball target detection frame in the picture based on a convolutional neural network, and performing model parameter optimization by using a back propagation algorithm to obtain a ball detection model.
2. The ball target detection is carried out by using a traditional method, in particular by using Harr and HOG characteristics combined with machine learning such as SVM and Adaboost methods.
In addition, a ball tracking model needs to be prepared in advance to always track the position of a ball target, and the method mainly comprises the following two steps:
1. learning the characteristics of the ball targets in the labeled video data by using a convolutional neural network in deep learning, extracting the ball characteristics from the current frame and the previous frame when in use, and comparing the similarity to predict the position change of the targets in the next frame;
2. and predicting the position of the target frame of the trackball in the next frame of image based on a Fourier transform and other realized method, a filtering method and a feature point-based optical flow method.
When the method is used, when balls appear in the range of the video acquisition equipment, a pre-trained ball detection method is started to detect a ball target frame, and then the ball target is tracked by using a target tracking method so as to always obtain the position and size information of the ball target in the detection frame of the current frame.
Specifically, the method for detecting the ball based on deep learning is to learn basketball labeling data by using a general target detection method to obtain a basketball detection model for detecting the ball; the traditional detection method is to detect the characteristics of balls, such as table tennis, basketball and football, by using Harr and HOG characteristic detection sub-combined with an SVM classification method, and is used for detecting the balls to obtain detection frames of the balls and the positions of the detection frames of the balls in a pixel coordinate system.
Action analytic system, this system need train sportsman's health key point detection model in advance, need use book machine neural network training to learn the picture of marking sportsman's health joint point, and the model that obtains sportsman's health joint point and detect is used for detecting sportsman's health key point.
During the use, combine the information that the sportsman who sportsman detecting system obtained detected the sportsman detection frame and detect the frame information to and the ball that ball detecting system obtained, operation sportsman's health joint point detection model obtains the information that sportsman's health joint point detected, synthesizes above-mentioned information and carries out sportsman's action judgement and analysis.
The action analysis system combines the information of the player detection frame obtained by the player detection system and the ball obtained by the ball detection systemDetecting frame information, predicting and calculating body joint points and limb positions of a player, wherein the prediction of the body joint points and the limb positions of the player is to obtain the position and type information of different joint points of a body in a pixel coordinate system by training the position information and the type information of the body joint points of the player based on a deep learning method, obtain the position information of related limbs according to the predicted position and type information of the joint points, and if the connecting line of the detected elbow and wrist points is the position information of the forearm, obtain the action information of the body by recording and combining the position information of the different joint points, integrate some preset action judging methods, and judge whether the player completes preset actions, such as high dribbling and low dribbling in a basketball, no ball agility training and bumping in a football, and judge the actions. The action judging method comprises the following steps: the completion of some motions relates to whether the difference between the position and type information such as the body joint point, the detection position frame of the ball and the like captured and detected by the method when the player makes motions and the information such as the key point, the detection position frame of the ball and the like set in the preset motion judgment method is less than a preset threshold value, and the expression formula is as follows:
n is the number of human joint points and the number of positions of ball detection frames, i is the ith joint point or detection frame, Xi is the ith predicted joint point position information or detection frame position information, Xoi is the ith preset joint point position information or detection frame position information, and τ is a preset threshold. If the value is less than the threshold value, the action is finished within a preset range, otherwise, the action is not finished. The combination of the actions can be richer through different joint points and different combination judgment of the positions of the detection frames, and further a more comprehensive training effect is achieved.
The visual feedback system is used for performing diversified interesting display feedback on a result of whether a golfer finishes a preset action in the action analysis system on display equipment, and the two most essential results which can appear are 'finished' and 'unfinished'. The system can perform richer visual effects and feedback and expansibility of certain sound effects, for example, if training contents of players perform low dribbling training at a set time, if the dribbling height is higher than a preset threshold value, the feedback system displays 'the dribbling height is reduced' and corresponding sound effects are sent out, if preset actions are met, prompts of 'finishing one low dribbling' and increasing points and sound effects in the training are displayed, and when the preset time is reached, the training is finished, the points are fed back, correct action playback in the training process is performed, and difference points of wrong action and correct action are played back.
Besides being applied to training, the system can be used as entertainment to set entertainment functions such as 'break through', and the like, due to the abundant visual feedback system and the expansibility for adding special effects, so that the training is more interesting, and the following example of the specific implementation mode is specifically shown.
When the system is applied, video acquisition equipment and visual terminal equipment need to be selected, in the example, a smart phone is selected and can be used as the acquisition equipment and the visual terminal equipment at the same time, and the APP of the smart phone is used as a using carrier of the system; for convenience and illustrative purposes, the example only uses part of the training movements in basketball training, but it can be generalized to a wider range of sports. This example is shown in fig. 2, and a single player uses the smartphone as a terminal to perform basketball left-hand single-hand low dribbling, left-hand and right-hand cross dribbling, dribbling contact training:
as shown in fig. 2, the training method of the present invention comprises the following steps:
(1) erecting a smart phone;
(2) adjusting the position of the player;
(3) the player makes a designated action;
(4) and (6) visualizing feedback.
In the step (1), when the smart phone is erected, as shown in fig. 3, 1 is the ground, 2 is the smart phone, 3 is a player, and 4 is a basketball, the illustration shows that the player turns on a front camera of the smart phone, fixes the smart phone on the ground or at other heights, and makes the screen of the smart phone shoot the body of the player;
in step (2), when the position of the player is adjusted, as shown in fig. 4, 1 is the player, 2 is the basketball, 3 is the detection frame preset by the mobile phone screen, 4 is the smart phone screen, and 5 is the position adjustment that the player needs to carry out: the detection frame of the player is completely appeared in a rectangular frame in the middle of the mobile phone screen only by moving the detection frame to the right relative to the smart phone. In the step, a player detection system is adopted to obtain the position and size information of the detection frame of the player so as to measure whether the used position is proper or not and to be used for a follow-up player motion analysis system.
In step (3), when the player makes a designated action, the action required by the screen needs to be completed, in this step, a specific basketball detection system in the ball detection system needs to be called to obtain the position and size information of the basketball detection frame, and a player action analysis system needs to be called to obtain the information of the body joint point and the limb position of the player, and whether the obtained information is consistent with the preset information of a certain action is compared, if so, the action is completed correctly, and if not, the action is not completed, in this example, the following actions are only shown and listed:
1. left-hand low-position racket ball: the deviation between the predicted body key point of the player and the position and type information of the basketball detection frame and the preset joint point information and the information of the basketball detection frame in continuous several frames of images must be smaller than a certain threshold value, and whether the basketball detection frame is a left-handed racket ball is judged according to the key point category information, if the deviation is judged according to the judgment method, the basketball detection frame position information and the body joint point information which are obtained by the ball detection system and the motion analysis system; if the height of the dribbling ball obtained by the position information of the joint point and the basketball hoop in the shooting process is lower than the preset relative height, the dribbling ball is shot in a low position, otherwise, the dribbling ball is not calculated;
2. basketball dribbling contact: the contact action is different from the action in that a certain mark is preset at a certain position in a screen, the mark is a mark which needs to be touched by hands or a mark which needs to be touched by a ball, and a player adjusts the actual position of the body or the position of the ball according to the relative distance and the direction of the player in the screen of the mobile phone and the mark so as to enable the detection frame of the body or the ball in the screen to be coincident with the position of the preset point in the screen; if the sum can be made, this point is touched, otherwise no touch is made. Specifically, the process can describe that the player needs to move according to the distance of seeing the point, which is relative to the screen, in the mobile phone by the player so as to realize that the difference between the joint point type and the position information of the body of the player and the position and the type information of the ball type and the position and the type information of the preset point is smaller than the preset threshold value, and then the player can be judged to successfully touch the preset point, as shown in fig. 5, 1 is the player, 2 is a basketball, 3 is a detection frame preset on a mobile phone screen, 4 is the mobile phone screen, 5 is a target needing to be touched by a left hand, 6 is the target needing to be touched by the basketball, 7 is the action of moving the left hand to the target, and 8 is the action of moving the basketball to the target. Furthermore, the black rectangle at the upper left corner in the figure is a preset range of points which need a player to touch with the left hand, if the player does not use the left hand when doing actions to touch, the player is regarded as an invalid action, and only the position of the left hand is coincided with the mark range through the movement of the player, the player is regarded as an effective action; the black triangle at the upper right corner is a preset range of points which need a player to touch with the ball, and the contact with the triangle point is regarded as an effective action only when the player carries the ball to the position. Such sports may exercise the stability and agility of basketball dribbles, and may also be used for dribbling exercises.
The visual feedback system comprises the following contents:
(1) the real-time player action feedback module is used for feeding back the result of whether each action of a player is correct or not in real time in the training process, calculating the difference between a preset point and a point where the player acts under the condition that the action is not standard, and feeding back the direction of the correct action;
(2) and the playback analysis module is finished, namely after the training of the module is finished, the player obtains the playback of each action and the difference between the body key point and the correct action key point, and calculates and analyzes the position where the correct action is adjusted according to the difference.
(3) The action scoring module, namely the system can score by integrating the conditions of correct action and wrong action of the training, provide more visual measuring standard for players and serve as the basis for the entertainment function to judge whether the next pass can be entered.
The invention can also be used for entertainment, and a specific application mode can be set to finish training actions with different difficulties and different times in a specified time, wherein the faster the finishing speed is, the higher the integral is, and whether to enter the next gate or not is judged according to the integral; or different action instructions can be set to randomly appear in one link to exercise the reaction capability of the player.
Claims (10)
1. The utility model provides a have ball or not training or amusement system based on artificial intelligence which characterized in that: the system comprises a player detection system, a ball detection system, an action analysis system and a visual feedback system;
when a player appears in the range of the view screen acquisition equipment, the player detection system learns player marking data in advance to obtain a player detection model, starts a player detection method through the model to obtain detection frame information of the player under a pixel coordinate system, and then tracks a player target by using a target tracking method, wherein the player detection method is a deep learning-based player detection method or a traditional detection method;
when balls appear in the acquisition range of acquisition equipment, the ball detection system learns ball labeling data in advance to obtain a ball detection model, starts a deep learning-based ball detection method or a traditional detection method through the model to obtain a ball detection frame and the position of the ball detection frame under a pixel coordinate system, and then tracks a ball target by using a target tracking method;
the motion analysis system is used for predicting and calculating body joint points and limb positions of the players and judging and analyzing the motions of the players by combining information of the player detection boxes obtained by the player detection system and information of the ball detection boxes obtained by the ball detection system;
and the visual feedback system displays and feeds back the result of whether the player finishes the preset action in the action analysis system on the display equipment.
2. The artificial intelligence based ball and non-ball training or entertainment system of claim 1, wherein: the player detection method is used for training data labeled with player target detection frames in pictures based on a convolutional neural network, and model parameters are optimized by using a back propagation method to obtain a player detection model.
3. The artificial intelligence based ball and non-ball training or entertainment system of claim 1, wherein: when a target tracking method is used for tracking the player target, the convolutional neural network in deep learning is utilized to learn the target characteristics of the players in the labeled video data, and the characteristics of the players are extracted from the current frame and the previous frame and compared with the similarity.
4. The artificial intelligence based ball and non-ball training or entertainment system of claim 1, wherein: when the player target is tracked by using a target tracking method, the position of a tracked player target frame in the next frame image is predicted based on a Fourier transform implementation method, a filtering method and a characteristic point-based optical flow method.
5. The artificial intelligence based ball and non-ball training or entertainment system of claim 1, wherein: in the action analysis system, the prediction of the body joint points and the limb positions of the players trains the position information and the type information of the body joint points of the players through a deep learning-based method to obtain the position and the type information of different joint points of the bodies in a pixel coordinate system, the position information of related limbs is obtained through the predicted position and type information of the joint points, and whether the players complete the preset actions is judged by combining a set action judgment method.
6. The artificial intelligence based ball and non-ball training or entertainment system of claim 5, wherein: the expression formula of the action judging method is as follows:
n is the number of human joint points and the number of positions of ball detection frames, i is the ith joint point or detection frame, Xi is the ith predicted joint point position information or detection frame position information, Xoi is the ith preset joint point position information or detection frame position information, and τ is a preset threshold.
7. The artificial intelligence based ball and non-ball training or entertainment system of claim 1, wherein: the visual feedback system comprises a real-time player action feedback module, an end playback analysis module and an action scoring module; the real-time player action feedback module feeds back a result of each action of the player in real time in the training process and feeds back a direction of correct action; the playback ending analysis module plays back the action after the training is ended; and the action scoring module is used for scoring the correct and wrong actions of the training.
8. The artificial intelligence based ball and non-ball training or entertainment system of claim 1, wherein: a plurality of display modes are preset in the visual feedback system, wherein the display mode is a 'breakthrough' mode formed when the score of the added action reaches a preset threshold value and enters the next training; the feedback is displayed as a visual special effect and a sound effect.
9. A method for training or entertainment with or without balls based on artificial intelligence is characterized in that: the method for an artificial intelligence based ball and non-ball training or entertainment system as claimed in claim 1, the method comprising the steps of:
(1) erecting a smart phone;
(2) adjusting the position of the player;
(3) the player makes a designated action;
(4) and (6) visualizing feedback.
10. The artificial intelligence based ball and non-ball training or entertainment method of claim 9, wherein: in the step (3), a detection model in the ball detection system is called to obtain the position and size information of a detection frame; and calling a player motion analysis system to obtain the information of body joint points and limb positions of the player, wherein the information is consistent with the information of a certain preset motion.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911080440.1A CN110929595A (en) | 2019-11-07 | 2019-11-07 | System and method for training or entertainment with or without ball based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911080440.1A CN110929595A (en) | 2019-11-07 | 2019-11-07 | System and method for training or entertainment with or without ball based on artificial intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110929595A true CN110929595A (en) | 2020-03-27 |
Family
ID=69853623
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911080440.1A Pending CN110929595A (en) | 2019-11-07 | 2019-11-07 | System and method for training or entertainment with or without ball based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110929595A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113095248A (en) * | 2021-04-19 | 2021-07-09 | 中国石油大学(华东) | Technical action correction method for badminton |
| DE102021000865A1 (en) | 2021-02-19 | 2021-11-18 | Felix Buchta | System and method for training safe meetings |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080182685A1 (en) * | 2001-09-12 | 2008-07-31 | Pillar Vision Corporation | Trajectory detection and feedback system for golf |
| CN108090421A (en) * | 2017-11-30 | 2018-05-29 | 睿视智觉（深圳）算法技术有限公司 | A kind of sportsman's competitive ability analysis method |
| CN108615055A (en) * | 2018-04-19 | 2018-10-02 | 咪咕动漫有限公司 | A kind of similarity calculating method, device and computer readable storage medium |
| CN109191588A (en) * | 2018-08-27 | 2019-01-11 | 百度在线网络技术（北京）有限公司 | Move teaching method, device, storage medium and electronic equipment |
-
2019
- 2019-11-07 CN CN201911080440.1A patent/CN110929595A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080182685A1 (en) * | 2001-09-12 | 2008-07-31 | Pillar Vision Corporation | Trajectory detection and feedback system for golf |
| CN108090421A (en) * | 2017-11-30 | 2018-05-29 | 睿视智觉（深圳）算法技术有限公司 | A kind of sportsman's competitive ability analysis method |
| CN108615055A (en) * | 2018-04-19 | 2018-10-02 | 咪咕动漫有限公司 | A kind of similarity calculating method, device and computer readable storage medium |
| CN109191588A (en) * | 2018-08-27 | 2019-01-11 | 百度在线网络技术（北京）有限公司 | Move teaching method, device, storage medium and electronic equipment |
Non-Patent Citations (1)
| Title |
|---|
| 张斌等: "篮球视频中基于AdaBoost分类器的运动员检测方法", 《湘潭大学自然科学学报》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102021000865A1 (en) | 2021-02-19 | 2021-11-18 | Felix Buchta | System and method for training safe meetings |
| CN113095248A (en) * | 2021-04-19 | 2021-07-09 | 中国石油大学(华东) | Technical action correction method for badminton |
| CN113095248B (en) * | 2021-04-19 | 2022-10-25 | 中国石油大学(华东) | Technical action correcting method for badminton |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11642047B2 (en) | Interactive training of body-eye coordination and reaction times using multiple mobile device cameras | |
| US10010778B2 (en) | Systems and methods for tracking dribbling and passing performance in sporting environments | |
| CN108564596B (en) | Intelligent comparison analysis system and method for golf swing video | |
| US10369445B2 (en) | Video analytics for human performance | |
| CN111444890A (en) | Sports data analysis system and method based on machine learning | |
| WO2019244153A1 (en) | Device, system, and method of computer vision, object tracking, image analysis, and trajectory estimation | |
| CN107803010B (en) | Table tennis training system | |
| CN107871120A (en) | Competitive sports based on machine learning understand system and method | |
| US20230372803A1 (en) | Tennis self-training system | |
| CN111259716A (en) | Human body running posture identification and analysis method and device based on computer vision | |
| US11941916B2 (en) | Practice drill-related features using quantitative, biomechanical-based analysis | |
| CN112487965A (en) | Intelligent fitness action guiding method based on 3D reconstruction | |
| CN110929595A (en) | System and method for training or entertainment with or without ball based on artificial intelligence | |
| CN116271766A (en) | Tennis training simulation method and device, electronic equipment and storage medium | |
| KR101864039B1 (en) | System for providing solution of justice on martial arts sports and analyzing bigdata using augmented reality, and Drive Method of the Same | |
| Huang et al. | Analysis Technology of Tennis Sports Match Based on Data Mining and Image Feature Retrieval | |
| CN112057833A (en) | Badminton forehand high-distance ball flapping motion identification method | |
| Malawski | Real-Time First Person Perspective Tracking and Feedback System for Weapon Practice Support in Fencing. | |
| Yang et al. | Implementing the tracking of 1500 m runners using Open CV | |
| US20230181971A1 (en) | System and method for analyzing one or movements of a moveable entity | |
| Wessa et al. | Can pose classification be used to teach Kickboxing? | |
| CN117475514A (en) | Shooting training system and method based on image analysis | |
| KR102623146B1 (en) | Apparatus and method for recommending golf clubs | |
| Shih | Aiming strategy error analysis and verification of a billiard training system | |
| Chinaglia et al. | Influence of Video Instruction on Football Kick Velocity in Young Players |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200327 |
|
| RJ01 | Rejection of invention patent application after publication |