CN112702481A - Table tennis track tracking device and method based on deep learning - Google Patents
Table tennis track tracking device and method based on deep learning Download PDFInfo
- Publication number
- CN112702481A CN112702481A CN202011369565.9A CN202011369565A CN112702481A CN 112702481 A CN112702481 A CN 112702481A CN 202011369565 A CN202011369565 A CN 202011369565A CN 112702481 A CN112702481 A CN 112702481A
- Authority
- CN
- China
- Prior art keywords
- table tennis
- ping
- pong
- camera
- processing unit
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 11
- 238000013135 deep learning Methods 0.000 title abstract description 6
- 238000012545 processing Methods 0.000 claims abstract description 28
- 238000001514 detection method Methods 0.000 claims description 18
- 230000005540 biological transmission Effects 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 11
- 238000013528 artificial neural network Methods 0.000 claims description 9
- 230000000007 visual effect Effects 0.000 claims description 8
- 238000001914 filtration Methods 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000001629 suppression Effects 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 4
- 230000006835 compression Effects 0.000 claims description 3
- 238000007906 compression Methods 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 3
- 238000003064 k means clustering Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 230000002194 synthesizing effect Effects 0.000 claims description 2
- 238000004458 analytical method Methods 0.000 abstract description 4
- 230000005055 memory storage Effects 0.000 abstract 1
- 238000013527 convolutional neural network Methods 0.000 description 4
- 238000012216 screening Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000003491 array Methods 0.000 description 2
- 238000012552 review Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000452 restraining effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 102220542003 tRNA (guanine-N(7)-)-methyltransferase_S27E_mutation Human genes 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/50—Constructional details
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Signal Processing (AREA)
- Multimedia (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a table tennis track tracking device and method based on deep learning. Two cameras catch the table tennis picture, use FPGA, vision processing unit, central processing unit to combine gesture sensing racket to obtain the table tennis orbit to the orbit analysis obtains the sportsman's batting condition, uses memory storage data, and uploads the high in the clouds through the GPRS module, realizes looking over in real time. The invention realizes real-time acquisition of the three-dimensional position of the table tennis and drawing of the motion trail of the table tennis, thereby analyzing the batting condition of players and realizing embedded deployment of equipment, and the invention has the advantages of small resource utilization and simple and convenient operation.
Description
Technical Field
The invention belongs to the embedded field, and particularly relates to a table tennis track tracking device and method based on deep learning.
Background
At present, common table tennis match events are analyzed through subjective judgment of human eyes and coaches, the coaches observe tracks, visual malposition exists on the table tennis positions due to the fact that the coaches are in single visual angles, and the coaches cannot accurately know the three-dimensional track information of the table tennis. Accurate event analysis places high demands on the ability of the coach, which also makes it impossible for the players to get a better shot plan in most table tennis events. In addition, in some cases, a coach needs to review a previous event video to review a table tennis game, and multiple switching of the camera view angle and a single view angle also affect judgment of the coach on a table tennis track and correct analysis of the coach on the event.
In the daily training of table tennis players, the table tennis players are difficult to objectively know the hitting track rules of the table tennis players, and find a correct method to correct the hitting defects of the table tennis players, so that the scientific analysis of the daily training of the table tennis players also becomes a great problem.
Disclosure of Invention
In view of the above, the present invention provides a table tennis ball tracking apparatus based on a neural network, including: posture sensing racket, six sensors, wireless transmission equipment, first camera, second camera, USB3.1 transmission equipment, FPGA, vision processing unit, central processing unit, display screen, memory, GPRS module and network equipment, wherein:
a posture sensor is arranged in the posture sensing racket to acquire a batting state;
the six-axis sensor is arranged on the first camera and the second camera to acquire the pitching angle of the cameras;
the wireless transmission equipment wirelessly transmits the data of the attitude sensing racket and the six-axis sensor to the central processing unit at a frequency band of 2.4 GHz;
the first camera and the second camera are two high-speed industrial cameras;
the FPGA carries out network input picture preprocessing, receives real-time images of the first camera and the second camera, and carries out size compression and filtering algorithm;
the visual processing unit has high DNN performance, acquires FPGA (field programmable gate array) preprocessed pictures, and operates a ping-pong target detection network based on RCNN (radar cross-correlation neural network) to acquire ping-pong identification results;
the central processing unit is used for acquiring a network identification result, receiving data transmitted by the wireless transmission equipment, synthesizing two-dimensional coordinates of the ping-pong, identifying the ping-pong table based on gradient descent, and analyzing to obtain batting data;
the memory stores the ping-pong ball track data and the batting data processed by the central processing unit;
the display screen displays the processing result of the central processing unit;
the GPRS module is connected with the display screen and uploads a table tennis identification track and batting data to the network equipment;
the network equipment is a mobile terminal capable of being networked and checks the uploading data of the GPRS module
Preferably, the first camera is horizontally arranged and is kept at the same horizontal position with a net of the table tennis table and is 0.37 meters away from the table tennis table, and the second camera is vertically arranged and is directly above the net of the table tennis table and above the center of the table tennis table and is 0.88 meters away from the table tennis table.
Based on the aim, the invention also provides a ping-pong ball trajectory tracking method based on the neural network, and the device comprises ping-pong ball table identification based on gradient descent and ping-pong ball target detection based on RCNN.
Preferably, the table tennis table identification based on gradient descent comprises the following steps:
s10, convolution noise reduction is carried out by using a Gaussian smoothing filter;
s20, calculating the gradient by using a first-order partial derivative operator,
gradient amplitude and corresponding direction calculation formula:
wherein G isxFor horizontal x-direction mask templates, GyIs a mask template in the vertical y direction, theta is a straight line angle,
s30, carrying out non-maximum value suppression and extracting the edge of the single pixel frame;
s40, obtaining picture edge information by using a double-threshold mode;
s50, creating a straight line family through the picture edge information point set, and discretizing a straight line angle theta to be-45 degrees, 0 degrees, 45 degrees and 90 degrees;
s60, determining a linear family length R as xcos θ + ysin θ according to the point coordinates (x, y) and the linear angle θ;
s70, taking a local maximum value according to the R value, and filtering interference straight lines through the shape and color information of the table tennis table;
s80, acquiring the edge straight line contour of the ping-pong table through the picture edge information;
and S90, calculating the intersection points of the four straight line contour coordinates to obtain the position information of the ping-pong table.
Preferably, the RCNN-based table tennis target detection comprises the following steps:
s11, setting the picture input size to 416 x 416, using CSPdark net53_ tiny as the main feature extraction network to extract the image feature, setting the feature layer to divide the picture x times according to the radius r of the ping-pong pixel,
obtaining a feature layer shape (x, x, 18);
s21, acquiring a data set, dividing the data set into a training set, a testing set and a verification set,
and S31, clustering the coordinates of the bounding box on the training set by adopting a K-means clustering algorithm, and calculating the coordinates of 3 bounding boxes of the single-scale convolutional layer feature map.
Compared with the prior art, the table tennis track tracking device based on deep learning disclosed by the invention has the following advantages:
1. the device has lower manufacturing cost and the equipment is easy to deploy;
2. using the RCNN-based table tennis target detection network, the table tennis position can be identified more quickly and accurately
Placing;
3. the vision processing unit is driven by the recognition module to reason the ping-pong target detection network, so that embedded application can be realized, and resource consumption is reduced.
Drawings
In order to make the object, technical scheme and beneficial effect of the invention more clear, the invention provides the following drawings for explanation:
fig. 1 is a schematic structural diagram of a table tennis ball trajectory tracking device according to an embodiment of the present invention;
fig. 2 is a block diagram of a table tennis ball trajectory tracking device according to an embodiment of the present invention.
Detailed Description
In view of the above, the present invention provides a table tennis ball tracking apparatus based on a neural network, including: posture sensing racket 14, six sensors 13, wireless transmission device 24, first camera 11, second camera 12, USB3.1 transmission device 21, FPGA22, visual processing unit 23, central processing unit 31, display screen 41, memory 32, GPRS module 33 and network device 42, wherein:
a posture sensor is arranged in the posture sensing racket 14 to obtain a batting state;
the six-axis sensor 13 is arranged on the first camera 11 and the second camera 12 and used for acquiring the pitching angle of the cameras;
the wireless transmission equipment 24 is used for wirelessly transmitting data of the attitude sensing racket 14 and the six-axis sensor 13 to the central processing unit 31 in a 2.4GHz frequency band;
the first camera 11 and the second camera 12 are two high-speed industrial cameras;
the FPGA22 is used for preprocessing network input pictures, receiving real-time images of the first camera 11 and the second camera 12 and performing size compression and filtering algorithms;
the visual processing unit 23 has high DNN performance, acquires FPGA22 preprocessed pictures, and runs a ping-pong target detection network based on RCNN to acquire ping-pong identification results; DNN is Deep Neural Networks;
the central processing unit 31 acquires a network identification result, receives data transmitted by the wireless transmission equipment, synthesizes two-dimensional coordinates of the ping-pong, identifies the ping-pong table based on gradient descent, and analyzes to obtain batting data;
a memory 32 for storing the ping-pong ball trajectory data and the batting data processed by the central processing unit 31;
a display screen 41 for displaying the processing result of the central processing unit 31;
the GPRS module 33 is connected with the display screen 41 and uploads the table tennis identification track and the hitting data to the network device 42;
the network device 42, which is a network-enabled mobile terminal, views the data uploaded by the GPRS module 33.
The first camera 11 is horizontally placed and kept at the same horizontal position with the net of the table tennis table, and is 0.37 meters away from the table tennis table, and the second camera 12 is vertically placed and is 0.88 meters away from the table tennis table directly above the net of the table tennis table and above the center of the table tennis table.
In particular embodiments, the attitude sensor includes MPU 9250; the six-axis sensor 13 includes an MPU 6050; the wireless transmission device 24 includes NRF24L 01; the models of the first camera 11 and the second camera 12 are MV-LD-8-3M-A, the first camera 11 and the second camera 12 are also connected with a camera, and the model of the camera body is MV-SUA50 GM-T; the model of the FPGA22 is ZynqUltraScale + MPSoc XCZU3EG-SFVC784-1-I Quad code-A531.2GHz; the central processor 31 comprises a CORE i 78700; memory 32 includes ST1000NM 0008; the display screen 41 includes S27E 360H; the GPRS module 33 comprises a SIM800A module.
The invention also provides a ping-pong ball trajectory tracking method based on the neural network, and the device comprises ping-pong ball table identification based on gradient descent and ping-pong ball target detection based on RCNN.
The identification of the ping-pong table based on gradient descent comprises the following steps:
s10, convolution noise reduction is carried out by using a Gaussian smoothing filter;
s20, calculating the gradient by using a first-order partial derivative operator,
gradient amplitude and corresponding direction calculation formula:
wherein G isxFor horizontal x-direction mask templates, GyIs a mask template in the vertical y direction, theta is a straight line angle,
s30, carrying out non-maximum value suppression and extracting the edge of the single pixel frame;
s40, obtaining picture edge information by using a double-threshold mode;
s50, creating a straight line family through the picture edge information point set, and discretizing a straight line angle theta to be-45 degrees, 0 degrees, 45 degrees and 90 degrees;
s60, determining a linear family length R as xcos θ + ysin θ according to the point coordinates (x, y) and the linear angle θ;
s70, taking a local maximum value according to the R value, and filtering interference straight lines through the shape and color information of the table tennis table;
s80, acquiring the edge straight line contour of the ping-pong table through the picture edge information;
and S90, calculating the intersection points of the four straight line contour coordinates to obtain the position information of the ping-pong table.
The RCNN-based table tennis target detection method comprises the following steps:
s11, setting the picture input size to 416 x 416, using CSPdark net53_ tiny as the main feature extraction network to extract the image feature, setting the feature layer to divide the picture x times according to the radius r of the ping-pong pixel,
obtaining a feature layer shape (x, x, 18);
s21, acquiring a data set, dividing the data set into a training set, a testing set and a verification set,
and S31, clustering the coordinates of the bounding box on the training set by adopting a K-means clustering algorithm, and calculating the coordinates of 3 bounding boxes of the single-scale convolutional layer feature map.
In a specific embodiment, the gradient-based ping-pong table identification comprises the following steps:
s10, carrying out convolution noise reduction and noise elimination on the input picture by using a Gaussian kernel with the size of 5;
s20, calculating the gradient by using a first-order partial derivative operator,
gradient amplitude and corresponding direction calculation formula:
Gxfor horizontal x-direction mask templates, GyIs a vertical y-direction mask template.
Calculating gradient amplitude G and a corresponding angle theta in the image;
s30, according to the image gradient, searching pixel points to carry out local maximum, setting the gray value corresponding to the non-maximum point as a background pixel point, judging the local optimum value of the pixel neighborhood region meeting the gradient value as the edge of the pixel, restraining the related information of the pixel and the non-maximum value, using the criterion to propose most non-edge pixels and extracting the edge of a single pixel frame;
s40, setting a high threshold and a low threshold, reserving pixel points of which the amplitudes of the pixel positions exceed the high threshold as edge pixels, and excluding pixel points of which the amplitudes are lower than the low threshold, wherein if the amplitudes of the pixel positions are between the two thresholds, the pixel is reserved only when the pixel is connected to a high-threshold pixel;
s50, setting the straight line angle theta to be-45 degrees, 0 degrees, 45 degrees and 90 degrees, and establishing a straight line set with the straight line angle theta for all the reserved pixel points;
s60, obtaining a linear family length R ═ xcos θ + ysin θ according to the point coordinates (x, y) and the linear angle θ, where each point corresponds to a straight line in the image space and is converted into polar coordinate parameters (R, θ);
s70, finding the most sinusoidal curve number of passing points (R, theta) in a polar coordinate parameter space to be set as a remaining straight line, calculating the slope k and the intercept b of each straight line, if the table tennis table is trapezoidal according to the shape characteristics of the table tennis table, screening the table tennis table into four straight line arrays of an upper straight line array, a lower straight line array, a left straight line array and a right straight line array through k and b, and then screening the table tennis table through edge color, if the table tennis table edge is white, setting an RGB value threshold value to enable only one straight line to be left in the four straight line arrays;
s80, determining four straight lines, and taking the received information as the straight line of the edge of the ping-pong table;
s90, calculating k and b values of four straight lines, taking any two points on the straight lines, and calculating intersection points of four points on the two straight lines through a determinant to obtain position information of four vertexes of the ping-pong table;
s100, compressing the sizes of the image 1 and the image 2 by using an image algorithm until the sizes meet the input size of a table tennis target detection network, and simultaneously performing a filtering preprocessing algorithm to eliminate noise of a high-speed camera, perform white balance operation and eliminate illumination influence.
The RCNN-based ping-pong target detection network is an algorithm for successfully applying deep learning to target detection, wherein RCNN is Region-CNN. The R-CNN realizes a target detection technology based on algorithms such as a Convolutional Neural Network (CNN), linear regression, a Support Vector Machine (SVM) and the like.
S11, in order to set the picture input size to 416 x 416, using CSPdark net53_ tiny as the main feature extraction network to extract the image feature, according to the ping-pong pixel radius r, setting the feature layer to divide the picture x times,
obtaining a feature layer shape (x, x, 18);
s21, selecting multi-scene conditions to record the ping-pong ball hitting video, extracting and screening pictures of the ping-pong ball video, setting the picture size according to network input, labeling training labels by using labellimg to obtain a data set, and then recording the data set according to the ratio of 3: 1 into a training set and a test set;
s31, clustering the coordinates of the bounding box on the data set by adopting a K-means mean value clustering algorithm, calculating the coordinates of 3 bounding boxes of the single scale convolutional layer characteristic diagram, and setting the coordinates as a ping-pong ball target detection network anchor box. And training for multiple times to obtain the target detection network.
Through a PCIE interface, IO is used for controlling a plurality of visual processing units to carry out asynchronous reasoning on a table tennis target detection network, reasoning data is returned to an identification module, and non-maximum suppression algorithm is used for screening out table tennis position information in two real-time images.
Through the combination of the ping-pong ball position information in the two visual angles and the ping-pong table position information, a three-dimensional coordinate system is established by using the ping-pong table, and the ping-pong balls are regarded as point coordinates in the coordinate system.
And repeatedly acquiring the coordinates of the table tennis points and drawing the coordinates as the table tennis tracks.
The present invention is not limited to the above preferred embodiments, and any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (5)
1. A table tennis ball trajectory tracking device based on a neural network is characterized by comprising: posture sensing racket, six sensors, wireless transmission equipment, first camera, second camera, USB3.1 transmission equipment, FPGA, vision processing unit, central processing unit, display screen, memory, GPRS module and network equipment, wherein:
a posture sensor is arranged in the posture sensing racket to acquire a batting state;
the six-axis sensor is arranged on the first camera and the second camera to acquire the pitching angle of the cameras;
the wireless transmission equipment wirelessly transmits the data of the attitude sensing racket and the six-axis sensor to the central processing unit at a frequency band of 2.4 GHz;
the first camera and the second camera are two high-speed industrial cameras;
the FPGA carries out network input picture preprocessing, receives real-time images of the first camera and the second camera, and carries out size compression and filtering algorithm;
the visual processing unit has high DNN performance, acquires FPGA (field programmable gate array) preprocessed pictures, and operates a ping-pong target detection network based on RCNN (radar cross-correlation neural network) to acquire ping-pong identification results;
the central processing unit is used for acquiring a network identification result, receiving data transmitted by the wireless transmission equipment, synthesizing two-dimensional coordinates of the ping-pong, identifying the ping-pong table based on gradient descent, and analyzing to obtain batting data;
the memory stores the ping-pong ball track data and the batting data processed by the central processing unit;
the display screen displays the processing result of the central processing unit;
the GPRS module is connected with the display screen and uploads a table tennis identification track and batting data to the network equipment;
the network equipment is a mobile terminal capable of being networked and checks the data uploaded by the GPRS module.
2. The apparatus of claim 1, wherein the first camera is positioned horizontally and is positioned 0.37 meters from the table tennis table at a level with the net of the table tennis table, and the second camera is positioned vertically and is positioned 0.88 meters from the table tennis table directly above the net of the table tennis table and above the center of the table tennis table.
3. A neural network-based ping-pong ball trajectory tracking method, characterized in that, by using the apparatus of claim 1 or 2, the method comprises gradient descent-based ping-pong table recognition and RCNN-based ping-pong target detection.
4. The method of claim 3, wherein the gradient descent based ping-pong table identification comprises the steps of:
s10, convolution noise reduction is carried out by using a Gaussian smoothing filter;
s20, calculating the gradient by using a first-order partial derivative operator,
gradient amplitude and corresponding direction calculation formula:
wherein G isxFor horizontal x-direction mask templates, GyIs a mask template in the vertical y direction, theta is a straight line angle,
s30, carrying out non-maximum value suppression and extracting the edge of the single pixel frame;
s40, obtaining picture edge information by using a double-threshold mode;
s50, creating a straight line family through the picture edge information point set, and discretizing a straight line angle theta to be-45 degrees, 0 degrees, 45 degrees and 90 degrees;
s60, determining a linear family length R as xcos θ + ysin θ according to the point coordinates (x, y) and the linear angle θ;
s70, taking a local maximum value according to the R value, and filtering interference straight lines through the shape and color information of the table tennis table;
s80, acquiring the edge straight line contour of the ping-pong table through the picture edge information;
and S90, calculating the intersection points of the four straight line contour coordinates to obtain the position information of the ping-pong table.
5. The method of claim 3, wherein the RCNN-based table tennis target detection comprises the steps of:
s11, setting the picture input size to 416 x 416, using CSPdark net53_ tiny as the main feature extraction network to extract the image feature, setting the feature layer to divide the picture x times according to the radius r of the ping-pong pixel,
obtaining a feature layer shape (x, x, 18);
s21, acquiring a data set, dividing the data set into a training set, a testing set and a verification set,
and S31, clustering the coordinates of the bounding box on the training set by adopting a K-means clustering algorithm, and calculating the coordinates of 3 bounding boxes of the single-scale convolutional layer feature map.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011369565.9A CN112702481B (en) | 2020-11-30 | 2020-11-30 | Table tennis track tracking device and method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011369565.9A CN112702481B (en) | 2020-11-30 | 2020-11-30 | Table tennis track tracking device and method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112702481A true CN112702481A (en) | 2021-04-23 |
| CN112702481B CN112702481B (en) | 2024-04-16 |
Family
ID=75506555
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011369565.9A Active CN112702481B (en) | 2020-11-30 | 2020-11-30 | Table tennis track tracking device and method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112702481B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114612522A (en) * | 2022-05-09 | 2022-06-10 | 广州精天信息科技股份有限公司 | Table tennis sport parameter detection method and device and table tennis training auxiliary system |
| CN115120949A (en) * | 2022-06-08 | 2022-09-30 | 乒乓动量机器人(昆山)有限公司 | Method, system and storage medium for realizing flexible batting strategy of table tennis robot |
| CN115624735A (en) * | 2022-10-12 | 2023-01-20 | 杭州欣禾圣世科技有限公司 | Auxiliary training system for ball games and working method |
| CN115835009A (en) * | 2022-11-30 | 2023-03-21 | 马鞍山师范高等专科学校 | Table tennis motion acquisition device embedded with acceleration gyroscope |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050031166A1 (en) * | 2003-05-29 | 2005-02-10 | Kikuo Fujimura | Visual tracking using depth data |
| CN101458434A (en) * | 2009-01-08 | 2009-06-17 | 浙江大学 | System for precision measuring and predicting table tennis track and system operation method |
| CN103019024A (en) * | 2012-11-29 | 2013-04-03 | 浙江大学 | System for realtime and accurate observation and analysis of table tennis rotating and system operating method |
| CN106651916A (en) * | 2016-12-29 | 2017-05-10 | 深圳市深网视界科技有限公司 | Target positioning tracking method and device |
| CN106780620A (en) * | 2016-11-28 | 2017-05-31 | 长安大学 | A kind of table tennis track identification positioning and tracking system and method |
| CN206853056U (en) * | 2017-01-24 | 2018-01-09 | 乐山师范学院 | Smart table tennis table |
-
2020
- 2020-11-30 CN CN202011369565.9A patent/CN112702481B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050031166A1 (en) * | 2003-05-29 | 2005-02-10 | Kikuo Fujimura | Visual tracking using depth data |
| CN101458434A (en) * | 2009-01-08 | 2009-06-17 | 浙江大学 | System for precision measuring and predicting table tennis track and system operation method |
| CN103019024A (en) * | 2012-11-29 | 2013-04-03 | 浙江大学 | System for realtime and accurate observation and analysis of table tennis rotating and system operating method |
| CN106780620A (en) * | 2016-11-28 | 2017-05-31 | 长安大学 | A kind of table tennis track identification positioning and tracking system and method |
| CN106651916A (en) * | 2016-12-29 | 2017-05-10 | 深圳市深网视界科技有限公司 | Target positioning tracking method and device |
| CN206853056U (en) * | 2017-01-24 | 2018-01-09 | 乐山师范学院 | Smart table tennis table |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114612522A (en) * | 2022-05-09 | 2022-06-10 | 广州精天信息科技股份有限公司 | Table tennis sport parameter detection method and device and table tennis training auxiliary system |
| CN114612522B (en) * | 2022-05-09 | 2023-01-17 | 广东金融学院 | Table tennis sport parameter detection method and device and table tennis training auxiliary system |
| CN115120949A (en) * | 2022-06-08 | 2022-09-30 | 乒乓动量机器人(昆山)有限公司 | Method, system and storage medium for realizing flexible batting strategy of table tennis robot |
| CN115120949B (en) * | 2022-06-08 | 2024-03-26 | 乒乓动量机器人(昆山)有限公司 | Method, system and storage medium for realizing flexible batting strategy of table tennis robot |
| CN115624735A (en) * | 2022-10-12 | 2023-01-20 | 杭州欣禾圣世科技有限公司 | Auxiliary training system for ball games and working method |
| CN115835009A (en) * | 2022-11-30 | 2023-03-21 | 马鞍山师范高等专科学校 | Table tennis motion acquisition device embedded with acceleration gyroscope |
| CN115835009B (en) * | 2022-11-30 | 2024-04-30 | 马鞍山师范高等专科学校 | Ping-pong ball motion acquisition device with embedded acceleration gyroscope |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112702481B (en) | 2024-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112702481B (en) | Table tennis track tracking device and method based on deep learning | |
| US11842572B2 (en) | Device, system, and method of computer vision, object tracking, image analysis, and trajectory estimation | |
| CN107301370B (en) | Kinect three-dimensional skeleton model-based limb action identification method | |
| Žemgulys et al. | Recognition of basketball referee signals from videos using Histogram of Oriented Gradients (HOG) and Support Vector Machine (SVM) | |
| CN107992881A (en) | A kind of Robotic Dynamic grasping means and system | |
| CN112154482B (en) | Ball game image analysis device and ball game image analysis method | |
| US20160307329A1 (en) | Robotic surveying of fruit plants | |
| CN109923583A (en) | A kind of recognition methods of posture, equipment and moveable platform | |
| CN105046206B (en) | Based on the pedestrian detection method and device for moving prior information in video | |
| CN106384106A (en) | Anti-fraud face recognition system based on 3D scanning | |
| WO2013101800A1 (en) | Camera calibration using feature identification | |
| JP2009531935A (en) | Apparatus, system, and method for determining compatibility with positioning instruction in person in image | |
| CN109359577B (en) | System for detecting number of people under complex background based on machine learning | |
| CN107153819A (en) | A kind of queue length automatic testing method and queue length control method | |
| CN110287907A (en) | A kind of method for checking object and device | |
| JP2006185109A (en) | Image measurement device and image measurement method | |
| US11790652B2 (en) | Detection of contacts among event participants | |
| CN112950673A (en) | Target object detection and tracking system and method | |
| CN114187664B (en) | Rope skipping counting system based on artificial intelligence | |
| CN108334870A (en) | The remote monitoring system of AR device data server states | |
| CN109389367A (en) | Staff attendance method, apparatus and storage medium | |
| Wang et al. | Deep learning-based human activity analysis for aerial images | |
| El-Sallam et al. | A low cost 3D markerless system for the reconstruction of athletic techniques | |
| CN114187663A (en) | Method for controlling unmanned aerial vehicle by posture based on radar detection gray level graph and neural network | |
| CN109895095B (en) | Training sample obtaining method and device and robot |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |