Recognizing Profile Faces by Imagining Frontal View

doi:10.1007/s11263-019-01252-7

	Recognizing Profile Faces by Imagining Frontal View
	Zhao, Jian 1; Xing, Junliang2 ; Xiong, Lin 3; Yan, Shuicheng 4,5; Feng, Jiashi 4
发表期刊	INTERNATIONAL JOURNAL OF COMPUTER VISION
ISSN	0920-5691
	2019-11-05
页码	19
通讯作者	Xing, Junliang(jlxing@nlpr.ia.ac.cn)
摘要	Extreme pose variation is one of the key obstacles to accurate face recognition in practice. Compared with current techniques for pose-invariant face recognition, which either expect pose invariance from hand-crafted features or data-driven deep learning solutions, or first normalize profile face images to frontal pose before feature extraction, we argue that it is more desirable to perform both tasks jointly to allow them to benefit from each other. To this end, we propose a Pose-Invariant Model (PIM) for face recognition in the wild, with three distinct novelties. First, PIM is a novel and unified deep architecture, containing a Face Frontalization sub-Net (FFN) and a Discriminative Learning sub-Net (DLN), which are jointly learned from end to end. Second, FFN is a well-designed dual-path Generative Adversarial Network which simultaneously perceives global structures and local details, incorporating an unsupervised cross-domain adversarial training and a meta-learning ("learning to learn") strategy using siamese discriminator with dynamic convolution for high-fidelity and identity-preserving frontal view synthesis. Third, DLN is a generic Convolutional Neural Network (CNN) for face recognition with our enforced cross-entropy optimization strategy for learning discriminative yet generalized feature representations with large intra-class affinity and inter-class separability. Qualitative and quantitative experiments on both controlled and in-the-wild benchmark datasets demonstrate the superiority of the proposed model over the state-of-the-arts.
关键词	Pose-invariant face recognition Face frontalization Cross-domain adversarial learning Meta-learning Learning to learn Enforced cross-entropy optimization Generative adversarial networks
DOI	10.1007/s11263-019-01252-7
关键词[WOS]	RECOGNITION
收录类别	SCI
语种	英语
资助项目	National Science Foundation of China[61672519] ; NUS startup[R-263-000-C08-133] ; NUS IDS[R-263-000-C67-646] ; ECRA[R-263-000-C87-133] ; MOE[R-263-000-C21-112] ; National Science Foundation of China[61672519] ; NUS startup[R-263-000-C08-133] ; NUS IDS[R-263-000-C67-646] ; ECRA[R-263-000-C87-133] ; MOE[R-263-000-C21-112]
项目资助者	National Science Foundation of China ; NUS startup ; NUS IDS ; ECRA ; MOE
WOS研究方向	Computer Science
WOS类目	Computer Science, Artificial Intelligence
WOS记录号	WOS:000494408100001
出版者	SPRINGER
七大方向——子方向分类	图像视频处理与分析
引用统计	被引频次：26[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/28873
专题	复杂系统认知与决策实验室_智能系统与工程
通讯作者	Xing, Junliang
作者单位	1.Inst North Elect Equipment, Beijing, Peoples R China 2.Chinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing, Peoples R China 3.JD Digits, Silicon Valley, CA USA 4.Natl Univ Singapore, Singapore, Singapore 5.Yitu Technol, Beijing, Peoples R China
通讯作者单位	模式识别国家重点实验室
推荐引用方式 GB/T 7714	Zhao, Jian,Xing, Junliang,Xiong, Lin,et al. Recognizing Profile Faces by Imagining Frontal View[J]. INTERNATIONAL JOURNAL OF COMPUTER VISION,2019:19.
APA	Zhao, Jian,Xing, Junliang,Xiong, Lin,Yan, Shuicheng,&Feng, Jiashi.(2019).Recognizing Profile Faces by Imagining Frontal View.INTERNATIONAL JOURNAL OF COMPUTER VISION,19.
MLA	Zhao, Jian,et al."Recognizing Profile Faces by Imagining Frontal View".INTERNATIONAL JOURNAL OF COMPUTER VISION (2019):19.