中文版 | English
Title

Face2Exp: Combating Data Biases for Facial Expression Recognition

Author
DOI
Publication Years
2022
Conference Name
2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
ISSN
1063-6919
ISBN
978-1-6654-6947-0
Source Title
Volume
2022-June
Pages
20259-20268
Conference Date
18-24 June 2022
Conference Place
New Orleans, LA, USA
Publication Place
10662 LOS VAQUEROS CIRCLE, PO BOX 3014, LOS ALAMITOS, CA 90720-1264 USA
Publisher
Abstract

Facial expression recognition (FER) is challenging due to the class imbalance caused by data collection. Existing studies tackle the data bias problem using only labeled facial expression dataset. Orthogonal to existing FER methods, we propose to utilize large unlabeled face recognition (FR) datasets to enhance FER. However, this raises another data bias problem—the distribution mismatch between FR and FER data. To combat the mismatch, we propose the Meta-Face2Exp framework, which consists of a base network and an adaptation network. The base network learns prior expression knowledge on class-balanced FER data while the adaptation network is trained to fit the pseudo labels of FR data generated by the base model. To combat the mismatch between FR and FER data, Meta-Face2Exp uses a circuit feedback mechanism, which improves the base network with the feedback from the adaptation network. Experiments show that our MetaFace2Exp achieves comparable accuracy to state-of-the-art FER methods with 10% of the labeled FER data utilized by the baselines. We also demonstrate that the circuit feedback mechanism successfully eliminates data bias.
 

Keywords
SUSTech Authorship
First
Language
English
URL[Source Record]
Indexed By
Funding Project
Guangdong Provincial Key Laboratory[2020B121201001] ; National Natural Science Foundation of China["62176170","62066042"]
WOS Research Area
Computer Science ; Imaging Science & Photographic Technology
WOS Subject
Computer Science, Artificial Intelligence ; Imaging Science & Photographic Technology
WOS Accession No
WOS:000870783006010
EI Accession Number
20224613119721
Data Source
IEEE
PDF urlhttps://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9879702
Citation statistics
Cited Times [WOS]:10
Document TypeConference paper
Identifierhttp://kc.sustech.edu.cn/handle/2SGJ60CL/406461
DepartmentDepartment of Computer Science and Engineering
工学院_深港微电子学院
工学院_斯发基斯可信自主研究院
Affiliation
1.Research Institue of Trustworthy Autonomous Systems, Southern University of Science and Technology & Department of Computer Science and Engineering, Southern University of Science and Technology
2.JD.com, Beijing, China
3.School of Microelectronics, Southern University of Science and Technology
4.Research Institue of Trustworthy Autonomous Systems, Southern University of Science and Technology & Department of Computer Science and Engineering, Southern University of Science and Technology
5.JD.com, Beijing, China
6.School of Microelectronics, Southern University of Science and Technology
First Author AffilicationDepartment of Computer Science and Engineering
First Author's First AffilicationDepartment of Computer Science and Engineering
Recommended Citation
GB/T 7714
Dan Zeng,Zhiyuan Lin,Xiao Yan,et al. Face2Exp: Combating Data Biases for Facial Expression Recognition[C]. 10662 LOS VAQUEROS CIRCLE, PO BOX 3014, LOS ALAMITOS, CA 90720-1264 USA:IEEE COMPUTER SOC,2022:20259-20268.
Files in This Item:
There are no files associated with this item.
Related Services
Fulltext link
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Export to Excel
Export to Csv
Altmetrics Score
Google Scholar
Similar articles in Google Scholar
[Dan Zeng]'s Articles
[Zhiyuan Lin]'s Articles
[Xiao Yan]'s Articles
Baidu Scholar
Similar articles in Baidu Scholar
[Dan Zeng]'s Articles
[Zhiyuan Lin]'s Articles
[Xiao Yan]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Dan Zeng]'s Articles
[Zhiyuan Lin]'s Articles
[Xiao Yan]'s Articles
Terms of Use
No data!
Social Bookmark/Share
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.