Learning view invariant gait features with Two-Stream GAN

doi:10.1016/j.neucom.2019.02.025

	Learning view invariant gait features with Two-Stream GAN
	Wang, Yanyun 1,2; Song, Chunfeng2,3 ; Huang, Yan2,3,4 ; Wang, Zhenyu 1; Wang, Liang2,3,4
刊名	NEUROCOMPUTING
	2019-04-28
卷号	339 页码:245-254
关键词	Gait recognition Cross-veiw Two-Stream GAN
ISSN号	0925-2312
DOI	10.1016/j.neucom.2019.02.025
通讯作者	Wang, Zhenyu(zywang@ncepu.edu.cn)
英文摘要	Gait recognition is an important yet challenging problem in computer vision. The changing view of gait is one of the most challenging factors, which could greatly affect the accuracy of cross-view gait recognition. In this paper, we propose a Two-Stream Generative Adversarial Network (TS-GAN) for cross-view gait recognition. For any view of gait representations, GAN can restore it to the corresponding standard view, to learn view invariant gait features. To achieve this goal, TS-GAN has two streams : (1) the global-stream can learn global contexts, and (2) the part-stream can learn local details. We combine the two streams to learn final identities. Moreover, we add a pixel-wise loss along with the generators of GAN to restore the gait details in pixel-level. We evaluate the proposed method on two widely used gait databases: CASIA-B and OU-ISIR. Experiment results show that our approach outperforms the compared state-of-the-art approaches. (C) 2019 Elsevier B.V. All rights reserved.
资助项目	National Key Research and Development Program of China[2016YFB10010 0 0] ; National National Science Foundation of China[61573139] ; National National Science Foundation of China[61525306] ; National National Science Foundation of China[61633021] ; National National Science Foundation of China[61721004] ; National National Science Foundation of China[61420106015] ; Capital Science and Technology Leading Talent Training Project[Z181100006318030] ; Fundamental Research Funds for the Central Universities[2018ZD05] ; Beijing Science and Technology Project[Z181100008918010]
WOS关键词	RECOGNITION ; REPRESENTATION ; IMAGE
WOS研究方向	Computer Science
语种	英语
出版者	ELSEVIER SCIENCE BV
WOS记录号	WOS:000461166500024
资助机构	National Key Research and Development Program of China ; National National Science Foundation of China ; Capital Science and Technology Leading Talent Training Project ; Fundamental Research Funds for the Central Universities ; Beijing Science and Technology Project
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/25004]
专题	中国科学院自动化研究所
通讯作者	Wang, Zhenyu
作者单位	1.North China Elect Power Univ, Sch Control & Comp Engn, Beijing 102206, Peoples R China 2.Chinese Acad Sci CASIA, CRIPAC, NLPR, Beijing 100190, Peoples R China 3.UCAS, Beijing 100190, Peoples R China 4.CEBSIT, Beijing 100190, Peoples R China
推荐引用方式 GB/T 7714	Wang, Yanyun,Song, Chunfeng,Huang, Yan,et al. Learning view invariant gait features with Two-Stream GAN[J]. NEUROCOMPUTING,2019,339:245-254.
APA	Wang, Yanyun,Song, Chunfeng,Huang, Yan,Wang, Zhenyu,&Wang, Liang.(2019).Learning view invariant gait features with Two-Stream GAN.NEUROCOMPUTING,339,245-254.
MLA	Wang, Yanyun,et al."Learning view invariant gait features with Two-Stream GAN".NEUROCOMPUTING 339(2019):245-254.