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

Face recognizing and aligning method based on shared convolutional characteristic

A technology of face detection and convolution, applied in the field of computer vision, can solve the problems of large amount of calculation of convolutional neural network, non-robustness of abnormal feature points and posture changes, and degradation of face recognition performance, so as to achieve performance improvement , strong robustness, and the effect of avoiding information loss

Active Publication Date: 2017-06-30
睿石网云(杭州)科技有限公司
View PDF6 Cites 15 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This method is not robust to abnormal feature points and pose changes, so it is difficult to get accurate face feature points, which will directly affect the face alignment effect and further lead to a serious decline in face recognition performance
[0005] Thirdly, although the method based on the convolutional neural network is higher than the traditional method in terms of accuracy, the convolutional neural network has a large amount of calculation and takes a long time to process a single image. It is difficult to meet the requirements of real-time face detection and alignment.

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 recognizing and aligning method based on shared convolutional characteristic
  • Face recognizing and aligning method based on shared convolutional characteristic
  • Face recognizing and aligning method based on shared convolutional characteristic

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0035] figure 1 The overall processing flow of the face detection and alignment method based on shared convolution features is given, and the present invention is further described below in conjunction with other drawings and specific implementation methods.

[0036] The present invention provides a face detection and alignment method based on shared convolution features. The main steps are as follows:

[0037] 1. Offline training steps

[0038] 1) Face detection

[0039] The first step is to train the face detection part first, and use the stochastic gradient descent method to figure 2The model shown is trained, the initial learning rate α is set to 0.01, and the total number of iterations is 80,000. After 50,000 iterations, the learning rate α is adjusted to 0.1 times the original value every 10,000 iterations. This method of gradually adjusting the learning rate is beneficial to the model. converge to a better solution.

[0040] 2) Facial feature localization

[0041...

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 face recognizing and aligning method based on a shared convolutional characteristic. The method comprises four steps of firstly performing multilayer convolution and pooling operation on an input image, and extracting a convolutional characteristic; performing face recognition according to the extracted convolutional characteristic, and outputting face position information in the image; then according to the face position information which is input in the previous step, extracting the convolutional characteristic which corresponds with the face position, performing face characteristic point positioning on the face, and outputting face characteristic point positions such as eyebrows, canthuses, nose and mouth corners in the face image; and performing rotating and scaling processing on the face according to a face image and corresponding face characteristic point points which are output in the face characteristic point detecting step, and outputting the aligned face image. The face recognizing and aligning method can realize automatic detection and automatic alignment for the face in the image and has advantages of high speed and high accuracy. Furthermore the method facilitates accuracy improvement of face recognition and face identification technology.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a face detection and alignment method based on shared convolution features. Background technique [0002] With the continuous development of computer science, human-computer interaction has become an increasingly important technology. As the face recognition and face verification technology in the field of computer vision, it has begun to be applied in the industry. In the past few decades, face recognition and face verification technology has been a hot research topic in the field of computer vision. Face detection and alignment is a crucial step in face recognition. [0003] First of all, the relatively common face detection method currently used is the face detection method based on Haar-like features and AdaBoost technology, which trains the face detector based on AdaBoost technology by extracting artificially designed features. However, because Haar-like features a...

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
CPCG06N3/04G06V40/164G06V40/171
Inventor 王华锋黄江刘万泉潘海侠
Owner 睿石网云(杭州)科技有限公司
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