FT-GAN: Face Transformation with Key Points Alignment for Pose-Invariant Face Recognition
Face recognition has been comprehensively studied. However, face recognition in the wild still suffers from unconstrained face directions. Frontal face synthesis is a popular solution, but some facial features are missed after synthesis. This paper presents a novel method for pose-invariant face rec...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2019-07-01
|
Series: | Electronics |
Subjects: | |
Online Access: | https://www.mdpi.com/2079-9292/8/7/807 |
id |
doaj-a34ac14103c34bb7a6dab2252a06e689 |
---|---|
record_format |
Article |
spelling |
doaj-a34ac14103c34bb7a6dab2252a06e6892020-11-24T21:29:17ZengMDPI AGElectronics2079-92922019-07-018780710.3390/electronics8070807electronics8070807FT-GAN: Face Transformation with Key Points Alignment for Pose-Invariant Face RecognitionWeiwei Zhuang0Liang Chen1Chaoqun Hong2Yuxin Liang3Keshou Wu4School of Computer and Information Engineering, Xiamen University of Technology, Xiamen 361024, ChinaSchool of Data and Computer Science, Sun Yat-Sen University, Guangzhou 510006, ChinaSchool of Computer and Information Engineering, Xiamen University of Technology, Xiamen 361024, ChinaSchool of Computer and Information Engineering, Xiamen University of Technology, Xiamen 361024, ChinaSchool of Computer and Information Engineering, Xiamen University of Technology, Xiamen 361024, ChinaFace recognition has been comprehensively studied. However, face recognition in the wild still suffers from unconstrained face directions. Frontal face synthesis is a popular solution, but some facial features are missed after synthesis. This paper presents a novel method for pose-invariant face recognition. It is based on face transformation with key points alignment based on generative adversarial networks (FT-GAN). In this method, we introduce CycleGAN for pixel transformation to achieve coarse face transformation results, and these results are refined by key point alignment. In this way, frontal face synthesis is modeled as a two-task process. The results of comprehensive experiments show the effectiveness of FT-GAN.https://www.mdpi.com/2079-9292/8/7/807face recognitionface pose transformationgenerative adversarial networkskey points alignment |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Weiwei Zhuang Liang Chen Chaoqun Hong Yuxin Liang Keshou Wu |
spellingShingle |
Weiwei Zhuang Liang Chen Chaoqun Hong Yuxin Liang Keshou Wu FT-GAN: Face Transformation with Key Points Alignment for Pose-Invariant Face Recognition Electronics face recognition face pose transformation generative adversarial networks key points alignment |
author_facet |
Weiwei Zhuang Liang Chen Chaoqun Hong Yuxin Liang Keshou Wu |
author_sort |
Weiwei Zhuang |
title |
FT-GAN: Face Transformation with Key Points Alignment for Pose-Invariant Face Recognition |
title_short |
FT-GAN: Face Transformation with Key Points Alignment for Pose-Invariant Face Recognition |
title_full |
FT-GAN: Face Transformation with Key Points Alignment for Pose-Invariant Face Recognition |
title_fullStr |
FT-GAN: Face Transformation with Key Points Alignment for Pose-Invariant Face Recognition |
title_full_unstemmed |
FT-GAN: Face Transformation with Key Points Alignment for Pose-Invariant Face Recognition |
title_sort |
ft-gan: face transformation with key points alignment for pose-invariant face recognition |
publisher |
MDPI AG |
series |
Electronics |
issn |
2079-9292 |
publishDate |
2019-07-01 |
description |
Face recognition has been comprehensively studied. However, face recognition in the wild still suffers from unconstrained face directions. Frontal face synthesis is a popular solution, but some facial features are missed after synthesis. This paper presents a novel method for pose-invariant face recognition. It is based on face transformation with key points alignment based on generative adversarial networks (FT-GAN). In this method, we introduce CycleGAN for pixel transformation to achieve coarse face transformation results, and these results are refined by key point alignment. In this way, frontal face synthesis is modeled as a two-task process. The results of comprehensive experiments show the effectiveness of FT-GAN. |
topic |
face recognition face pose transformation generative adversarial networks key points alignment |
url |
https://www.mdpi.com/2079-9292/8/7/807 |
work_keys_str_mv |
AT weiweizhuang ftganfacetransformationwithkeypointsalignmentforposeinvariantfacerecognition AT liangchen ftganfacetransformationwithkeypointsalignmentforposeinvariantfacerecognition AT chaoqunhong ftganfacetransformationwithkeypointsalignmentforposeinvariantfacerecognition AT yuxinliang ftganfacetransformationwithkeypointsalignmentforposeinvariantfacerecognition AT keshouwu ftganfacetransformationwithkeypointsalignmentforposeinvariantfacerecognition |
_version_ |
1725966335444254720 |