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

Face recognition method based on adversarial deep learning network

A network and face occlusion technology, applied in character and pattern recognition, biological neural network models, instruments, etc., can solve problems such as inaccurate face occlusion recognition

Inactive Publication Date: 2019-07-05
CENT SOUTH UNIV
View PDF0 Cites 25 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] 3. Strict requirements on lighting conditions
Image generation technology based on Generative Adversarial Nets (GAN) algorithm to solve the problem of inaccurate face occlusion recognition

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
  • Face recognition method based on adversarial deep learning network
  • Face recognition method based on adversarial deep learning network
  • Face recognition method based on adversarial deep learning network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0012] First, according to the Fast R-CNN target detection method, the selective search method is used to extract face area suggestions from the face data set UMDfaces, and the Fast R-CNN is applied to face detection to further integrate human The face data set is combined with the Faster R-CNN target detection method to realize face detection.

[0013] Secondly, in the Fast R-CNN-based generative confrontation network experiment, the Fast R-CNN was combined with the generative confrontation network to realize the detection of faces under occlusion. Inspired by the independent training of Faster R-CNN, the RPN (RegionProposal Network) replaces the selective search method to extract face image region proposals, and combines it with the FastR-CNN-based generative confrontation network for training. In the course of the experiment, it is finally concluded that the accuracy of the RPN network combined with the Fast R-CNN-based generative confrontation network in face occlusion det...

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 patent designs a CNN and generative adversarial network combined network framework, realizing detection of the human face under the condition of shielding. In recent years, the deep convolutionalneural network (CNN) significantly improves the accuracy of image classification and target detection. The invention aims to apply a method of combining target detection and an adversarial network todetection of a human face under a shielding condition. Although face recognition is applied to many occasions, most of current face recognition systems can only be applied to some strict and standardlimited environments, for example, when a detected main body is detected, posture adjustment needs to be carried out, and glasses or masks cannot be worn. In the method, the generative adversarial network is mainly used for data enhancement, and learning is carried out by generating a shielding difficulty case. According to the method, the original detector and confrontation are learned in a common mode, the detection performance is enhanced, and the precision is improved on the test result.

Description

technical field [0001] The invention aims to solve the environmental restriction problem of a face recognition system, improve the accuracy of face recognition, and belongs to the field of artificial intelligence machine learning. Background technique [0002] Machine learning is one of the branches of artificial intelligence, and it is at the core. As the name suggests, the research of machine learning aims to make computers learn to learn, to simulate human learning behavior, to build learning ability, and to realize recognition and judgment. Machine learning uses algorithms to analyze massive amounts of data, find patterns from them, complete learning, and use the learned thinking model to make decisions and predict real events. In short, machine learning is a method of artificial intelligence, deep learning is a technique for implementing machine learning, and generative adversarial networks are a classification in deep learning. The use of machine learning for face re...

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): G06K9/00G06N3/04
CPCG06V40/172G06N3/045
Inventor 周霞霞
Owner CENT SOUTH UNIV
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

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com