Gesture recognition method based on deep neural network and attention mechanism

A deep neural network and gesture recognition technology, applied in the field of gesture recognition, can solve the problems of limited interaction scenarios, large amount of calculation, expensive equipment, etc.

Pending Publication Date: 2021-09-10
BEIJING UNIV OF TECH
View PDF0 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] In order to solve the problems of expensive equipment, limited interaction scenarios, and large amount of calculation required by traditional gesture recognition methods, the present invention considers combining Efficient Channel Attention (ECA) and Single Shot MultiBox Detector (Single Shot MultiBox Detector, SSD) improves the dual-stream algorithm, and then establishes a more general vision-based gesture recognition model

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 deep neural network and attention mechanism
  • Gesture recognition method based on deep neural network and attention mechanism
  • Gesture recognition method based on deep neural network and attention mechanism

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0048] Concrete implementation of the present invention is divided into following four steps:

[0049] 1) Data loading and dual-stream algorithm implementation

[0050] 2) Implementation of SSD hand posture detection network

[0051] 3) Overall training in the public data set

[0052] 4) Analysis of experimental results

[0053] (1) Data loading and dual-stream network implementation

[0054] In this method, an average of 3 video frames are selected from the gesture video data as the input of the dual-stream algorithm spatial convolutional network, and then the optical flow data are extracted from the left and right of the 3 video frames and each of the 5 neighboring frames as the dual-stream algorithm time volume. input to the product network. In addition, in order to enhance the generalization of the gesture recognition method, this method randomly crops gesture frames, and each gesture frame needs to be pre-cut to a resolution size of 512×512.

[0055] Both the spatial...

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

A gesture recognition method based on a deep neural network and an attention mechanism belongs to the field of electronic information. The method comprises the following steps: firstly, introducing an ECA effective channel attention in a double-flow algorithm to enhance the attention of the double-flow algorithm on a gesture key frame, and respectively extracting space and time sequence characteristics in a dynamic gesture by utilizing a space convolutional network and a time convolutional network in the double-flow algorithm; secondly, a gesture frame with the highest attention degree is selected from the spatial stream through ECA, and corresponding hand posture features are extracted through the single-shot multi-frame detector technology; and finally, fusing the hand posture features with human body posture features and gesture time sequence features extracted from double flows, and then classifying and recognizing gestures. According to the method, verification is carried out on a Challarn2013 multi-mode gesture data set, the accuracy rate is 66.23%, and compared with a previous method that double-flow recognition is carried out on the data set only through RGB information, a better gesture recognition effect is obtained.

Description

technical field [0001] The invention belongs to the field of electronic information, and is a gesture recognition method based on a deep neural network and an attention mechanism, through which gesture video data captured by a common camera can be classified into corresponding text meanings. Background technique [0002] Gesture is an important part of human communication, and it is also an important way of human computer interaction (Human Computer Interaction, HCI). By detecting human gestures, it can help machines better understand human instructions, and then complete corresponding auxiliary tasks. For example, in a smart home environment, you can control the switch of the air conditioner and switch the TV screen through gestures; in the process of smart driving, you can also control some functions inside the car through gestures, so that the driver can focus more on the road. itself, reducing the occurrence of traffic accidents. [0003] At present, the research on ge...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/62G06N3/04G06F3/01
CPCG06F3/017G06N3/045G06F18/253
Inventor 何坚刘炎
Owner BEIJING UNIV OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products