Contrastive attention network with dense field estimation for face completion

doi:10.1016/j.patcog.2021.108465

CORC > 自动化研究所 > 中国科学院自动化研究所 > 智能感知与计算研究中心

	Contrastive attention network with dense field estimation for face completion
	Ma, Xin 1,2,5,6; Zhou, Xiaoqiang 1,2,4,6; Huang, Huaibo 1,2,5,6; Jia, Gengyun 1,2,5,6; Chai, Zhenhua 3; Wei, Xiaolin 3
刊名	PATTERN RECOGNITION
	2022-04-01
卷号	124 页码:13
关键词	Face completion Unsupervised learning Attention mechanism 3D Face analysis
ISSN号	0031-3203
DOI	10.1016/j.patcog.2021.108465
通讯作者	Ma, Xin(xin.ma@cripac.ia.ac.cn)
英文摘要	Most modern face completion approaches adopt an autoencoder or its variants to restore missing regions in face images. Encoders are often utilized to learn powerful representations that play an important role in meeting the challenges of sophisticated learning tasks. Specifically, various kinds of masks are often presented in face images in the wild, forming complex patterns, especially in this hard period of COVID19. It's difficult for encoders to capture such powerful representations under this complex situation. To address this challenge, we propose a self-supervised Siamese inference network to improve the generalization and robustness of encoders. It can encode contextual semantics from full-resolution images and obtain more discriminative representations. To deal with geometric variations of face images, a dense correspondence field is integrated into the network. We further propose a multi-scale decoder with a novel dual attention fusion module (DAF), which can combine the restored and known regions in an adaptive manner. This multi-scale architecture is beneficial for the decoder to utilize discriminative representations learned from encoders into images. Extensive experiments clearly demonstrate that the proposed approach not only achieves more appealing results compared with state-of-the-art methods but also improves the performance of masked face recognition dramatically. (c) 2021 Elsevier Ltd. All rights reserved.
资助项目	National Natural Science Foundatio of China[62006228]
WOS关键词	ADVERSARIAL NETWORK ; IMAGE
WOS研究方向	Computer Science ; Engineering
语种	英语
出版者	ELSEVIER SCI LTD
WOS记录号	WOS:000736980400001
资助机构	National Natural Science Foundatio of China
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/47139]
专题	自动化研究所_智能感知与计算研究中心
通讯作者	Ma, Xin
作者单位	1.CEBSIT, Beijing, Peoples R China 2.CASIA, CRIPAC, Beijing, Peoples R China 3.Visual Intelligence Dept, Meituan, Peoples R China 4.Univ Sci & Technol China, Beijing, Peoples R China 5.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing, Peoples R China 6.NLPR, Beijing, Peoples R China
推荐引用方式 GB/T 7714	Ma, Xin,Zhou, Xiaoqiang,Huang, Huaibo,et al. Contrastive attention network with dense field estimation for face completion[J]. PATTERN RECOGNITION,2022,124:13.
APA	Ma, Xin,Zhou, Xiaoqiang,Huang, Huaibo,Jia, Gengyun,Chai, Zhenhua,&Wei, Xiaolin.(2022).Contrastive attention network with dense field estimation for face completion.PATTERN RECOGNITION,124,13.
MLA	Ma, Xin,et al."Contrastive attention network with dense field estimation for face completion".PATTERN RECOGNITION 124(2022):13.