Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Gesture recognition method based on recurrent 3D convolutional neural network

A neural network and three-dimensional convolution technology, applied in the field of human-computer intelligent interaction, can solve the problems of difficult classification and achieve the effect of low equipment cost and convenient non-contact gesture recognition

Inactive Publication Date: 2018-01-16
BEIJING UNION UNIVERSITY
View PDF4 Cites 17 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This application is based on electromyographic signals for gesture recognition. Due to the complexity of the cluster structure and the influence of individual differences in electromyographic signals, electrode positions, etc., its classification is very difficult.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Gesture recognition method based on recurrent 3D convolutional neural network
  • Gesture recognition method based on recurrent 3D convolutional neural network
  • Gesture recognition method based on recurrent 3D convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0053] Such as figure 1 As shown, step 100 is executed to perform data preprocessing, which is used to process the acquired data into a fixed size, which meets the specification requirements of the input layer of the cyclic three-dimensional convolutional neural network. Such as figure 2 As shown, step 200 and step 210 are executed in sequence to crop the input data, and the random cropping size is A×A (in this embodiment, A=112 is set). Step 220 is executed, in order to increase the diversity of training samples, data enhancement is required. Carry out random spatial rotation and scaling to each video, the angle of spatial rotation is ± B ° (in this embodiment, setting B=15), zoom size is ± C% (in this embodiment, setting C=20) . Execute step 230, carry out random time scaling and jitter to each video, zoom size is ± D% (in this embodiment setting D=20), shaking amplitude is ± E frame (in this embodiment setting E= 3). Execute step 240 to obtain data meeting the input s...

Embodiment 2

[0060] Such as Figure 5 As shown, the overall system architecture consists of four parts: a data input module 500 , a data preprocessing module 510 , a recurrent 3D convolutional neural network classifier 520 and an output class label 530 . The loop three-dimensional convolutional neural network classifier 520 can be decomposed into: the loop three-dimensional convolutional neural network classifier design submodule 521, the loop three-dimensional convolutional neural network classifier pre-training submodule 522, the loop three-dimensional convolutional neural network classifier training submodule Module 523 , recurrent three-dimensional convolutional neural network classifier optimization submodule 524 and testing submodule 525 .

[0061] This embodiment proposes a method for gesture recognition based on a cyclic three-dimensional convolutional neural network, including importing video data in the data input module 500, performing data preprocessing in the data preprocessin...

Embodiment 3

[0063] Such as Image 6 As shown, step 600 is executed to collect image data through a camera (such as Figure 6a shown). Execute step 620, carry out gesture clipping to the collected data, remove redundant part, segment gesture image (such as Figure 6b shown). The image is processed in two stages, which are the training stage of the 3D convolutional neural network model and the testing stage of the 3D convolutional neural network model. In the training phase of the cyclic three-dimensional convolutional neural network model, step 620 is performed, and the data is enhanced by using data enhancement techniques, and the processing results are as follows: Figure 6c shown. Step 630 is executed to preprocess the data to extract clear key frames. Execute step 640, train the 3D convolutional neural network first, and then train the overall model. Such as Figure 6d As shown, the training method is to crop the video to obtain several pictures, and the size of random cropping ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention provides a gesture recognition method based on a recurrent 3D convolutional neural network, which includes a data preprocessing step. The method further includes the following steps: designing a recurrent 3D convolutional neural network classifier; pre-training the recurrent 3D convolutional neural network classifier; training the recurrent 3D convolutional neural network classifier;optimizing the recurrent 3D convolutional neural network classifier; outputting a classification label; and testing the classification effect of the classifier. Through the method, the effect of gesture recognition is more precise, the gestures in SKIG and ChaLearn2014 data sets can be accurately recognized, and the accuracy is higher than that of the traditional methods.

Description

technical field [0001] The invention relates to the technical field of human-computer intelligent interaction, in particular to a gesture recognition method based on a circular three-dimensional convolutional neural network. Background technique [0002] Gesture recognition is a topic in computer science and language technology. The purpose is to enable computers to recognize human gestures through mathematical algorithms, and to achieve effective communication and control between humans and computers. Gesture, with its advantages of intuitive, natural and easy-to-learn human-computer interaction and the realization of intelligent information exchange and processing without special hardware cooperation, has attracted more and more attention from enterprises and scholars. Among them, the current application Gesture recognition in the field of unmanned driving and VR is a simple gesture recognition. It is an artificial intelligence system that can understand the user's gestur...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06K9/00G06K9/62
Inventor 袁家政刘宏哲邓智方
Owner BEIJING UNION UNIVERSITY
Features
  • Generate Ideas
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com