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

Expression recognition method based on double-branch mixed residual connection

A facial expression recognition and dual-branch technology, applied in the field of image processing, can solve the problems of insufficient recognition robustness, large model, and reduced model parameters, etc., and achieve the effect of solving the recognition accuracy

Pending Publication Date: 2022-08-05
SOUTHWEAT UNIV OF SCI & TECH
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve the problems that there are few reliable expression recognition databases in the current expression recognition tasks, the recognition is not robust enough under uncontrolled conditions, and the deep learning to build an expression recognition model takes a long time to train and the model is relatively large. The facial expression recognition data set FER2013 was used for training and testing. Through the method described, a better recognition rate than existing algorithms was obtained when using a single network for training and without using additional training data.
[0005] In order to achieve the above object, the present invention provides an expression recognition method with dual-branch hybrid residual connection to solve the problems of recognition accuracy of expression recognition under uncontrolled conditions and the lightweight of the model, which mainly includes five parts , the first part is to enhance the basic image data; the second part is to construct a dual-branch hybrid residual network for feature extraction; the third part is to add an adaptive feature fusion module; the fourth part is to fuse depth separable convolution to reduce the amount of model parameters ; The fifth part is network training and testing, outputting the final expression classification prediction

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
  • Expression recognition method based on double-branch mixed residual connection
  • Expression recognition method based on double-branch mixed residual connection
  • Expression recognition method based on double-branch mixed residual connection

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0024] In order to better understand the present invention, the feature extraction network combining adaptive feature fusion and dual-branch hybrid residual connection feature extraction network of the present invention will be described in more detail below with reference to specific embodiments. In the following description, detailed descriptions of the current prior art may dilute the subject matter of the present invention, and such descriptions will be omitted here.

[0025] figure 1 It is a specific network model diagram of the present invention combining adaptive feature fusion and dual-branch mixed residual connection feature extraction network. The MRS module is a designed dual-branch mixed residual module with fusion depth and separable convolution. In this embodiment, the following steps are followed:

[0026] Step 1: Randomly zoom and crop the picture, and then flip the picture data horizontally to increase the picture database;

[0027] Step 2: Use mixup image f...

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 an expression recognition method based on double-branch mixed residual connection. The method specifically comprises the steps that ResNet18 is used as a trunk feature extraction network, a double-branch multi-residual connection mode is designed on the basis, and feature extraction is completed; secondly, comprehensively analyzing problems in two aspects of network depth and width, fusing depth separable convolution, and constructing a lightweight expression recognition network; and finally, according to the characteristics of the facial expression information, in order to obtain more accurate and perfect feature information, adding an adaptive feature fusion module ASFF, carrying out analysis and verification on the fused information, selecting a hierarchy with the most superior expression for output, and carrying out final expression classification prediction. According to the method, a perfect feature extraction mechanism and a self-adaptive feature fusion mode are combined, and common convolution is replaced by deep separable convolution, so that the problems of recognition precision of expression recognition under an uncontrolled condition and light weight of a model can be solved.

Description

technical field [0001] The present invention relates to image processing technology, in particular, to an expression recognition method of dual-branch mixed residual connection. The method combines adaptive feature fusion to construct a robust feature extraction network to obtain better facial expressions. identify. Background technique [0002] Facial expressions occupy a large proportion in the communication between people. Facial expression recognition is a technology in which computers try to understand human emotions by analyzing face information, and it has become a hot topic in the field of computer vision. With the development and deepening of facial expression research, the methods of expression recognition are becoming more and more diverse. To improve the accuracy and speed of facial expression recognition, as well as the various influencing factors associated with the recognition, are affected by the development of computer technology to a certain extent. Rese...

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): G06V40/16G06V10/80G06V10/82G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06F18/241G06F18/253
Inventor 张红英卢琇雯韩雪张奇
Owner SOUTHWEAT UNIV OF SCI & TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products