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

Single-sample face recognition method based on feature expansion

A single-sample, sample-person technology, applied in the field of face recognition, can solve problems such as lack of intra-class differences, inability to predict intra-class changes in test images, and affect face recognition performance

Active Publication Date: 2019-08-06
成都电科智达科技有限公司
View PDF11 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Since there is only one training sample per class, the lack of intra-class differences makes it impossible to predict the intra-class variation of test images, which seriously affects the performance of single-sample face 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
  • Single-sample face recognition method based on feature expansion
  • Single-sample face recognition method based on feature expansion
  • Single-sample face recognition method based on feature expansion

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach

[0041] The specific implementation is as follows:

[0042] A. Preprocess all images (mainly including face detection alignment and normalization)

[0043] B. Pre-training the classification model on the diverse face dataset CASIA-WebFace

[0044] C. Apply the pre-trained model to the single-sample training set, and extract the face features of the single-sample training set.

[0045] D. Expand the single-sample features in the feature space, and use the expanded features to fine-tune the last layer of softmax classifier.

[0046] E. Input the test data set into the trained network to get the recognition result

[0047] In the present invention, a single-sample training set recognition test is carried out on three data sets of ORL, LFW, and FERET. Among them, ORL has a total of 40 people, and each person has 10 face images, and each person chooses one as training, and the rest as testing. Select the first 50 people from the LFW data set with more than 10 samples for trainin...

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 belongs to the technical field of face recognition, and relates to a single-sample face recognition method based on feature expansion. Based on transfer learning, a deep convolutional neural network is adopted to extract face features with robustness, and a sample expansion method of a feature space is provided and comprises the following steps: firstly, based on transfer learning, training a deep convolutional neural network on a multi-sample public face set, applying the deep convolutional neural network to a target face data set, and extracting the face features by using a pre-trained model; and expanding the data in the feature space by using the intra-class difference of the auxiliary data set, and training a classifier by using the expanded data to obtain better identification performance. The sample expansion method based on the feature space overcomes the problem of insufficient samples, is more potential than the data enhancement of a common image domain, and improves the recognition rate of the model.

Description

technical field [0001] The invention belongs to the technical field of face recognition, and relates to a single-sample face recognition method based on feature expansion. Background technique [0002] Face recognition has become a popular research direction due to the fact that face recognition has become more and more widely used in practice. A large number of face recognition algorithms have been proposed, but most algorithms need sufficient representative training data to achieve good performance. In fact, it is very difficult to collect a large number of training samples, which is also the current face recognition technology. One of the main challenges. In some special occasions, such as law enforcement, passport verification, identity verification, etc., each person can only get one image. Especially in large-scale recognition applications, if you want to collect more training samples for each person, it will inevitably cause extremely expensive costs, so you can onl...

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): G06K9/00G06K9/62
CPCG06V40/168G06V40/172G06F18/214Y02T10/40
Inventor 曹宗杰徐升平
Owner 成都电科智达科技有限公司
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

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