Learning to Align via Wasserstein for Person Re-Identification

doi:10.1109/TIP.2020.2998931

CORC > 自动化研究所 > 中国科学院自动化研究所 > 精密感知与控制研究中心 > 人工智能与机器学习

	Learning to Align via Wasserstein for Person Re-Identification
	Zhang, Zhizhong 2,3; Xie, Yuan 4; Li, Ding 2,3; Zhang, Wensheng 2,3; Tian, Qi 1
刊名	IEEE TRANSACTIONS ON IMAGE PROCESSING
	2020
卷号	29 页码:7104-7116
关键词	Semantics Heating systems Measurement Learning systems Training Estimation Feature extraction Person re-identification deep metric learning convolutional neural network Wasserstein distance
ISSN号	1057-7149
DOI	10.1109/TIP.2020.2998931
通讯作者	Zhang, Wensheng(zhangwenshengia@hotmail.com) ; Tian, Qi(tian.qi1@huawei.com)
英文摘要	Existing successful person re-identification (Re-ID) models often employ the part-level representation to extract the fine-grained information, but commonly use the loss that is particularly designed for global features, ignoring the relationship between semantic parts. In this paper, we present a novel triplet loss that emphasizes the salient parts and also takes the consideration of alignment. This loss is based on the crossing-bing matching metric that also known as Wasserstein Distance. It measures how much effort it would take to move the embeddings of local features to align two distributions, such that it is able to find an optimal transport matrix to re-weight the distance of different local parts. The distributions in support of local parts is produced via a new attention mechanism, which is calculated by the inner product between high-level global feature and local features, representing the importance of different semantic parts w.r.t. identification. We show that the obtained optimal transport matrix can not only distinguish the relevant and misleading parts, and hence assign different weights to them, but also rectify the original distance according to the learned distributions, resulting in an elegant solution for the mis-alignment issue. Besides, the proposed method is easily implemented in most Re-ID learning system with end-to-end training style, and can obviously improve their performance. Extensive experiments and comparisons with recent Re-ID methods manifest the competitive performance of our method.
资助项目	National Key Research and Development Program of China[2017YFC0803700] ; National Natural Science Foundation of China[U1636220] ; National Natural Science Foundation of China[61772524] ; National Natural Science Foundation of China[61961160707] ; National Natural Science Foundation of China[61976212] ; Beijing Municipal Natural Science Foundation[4182067] ; Fundamental Research Funds for the Central Universities ; Shanghai Key Laboratory of Trustworthy Computing
WOS关键词	NETWORK
WOS研究方向	Computer Science ; Engineering
语种	英语
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
WOS记录号	WOS:000546910100015
资助机构	National Key Research and Development Program of China ; National Natural Science Foundation of China ; Beijing Municipal Natural Science Foundation ; Fundamental Research Funds for the Central Universities ; Shanghai Key Laboratory of Trustworthy Computing
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/40047]
专题	精密感知与控制研究中心_人工智能与机器学习
通讯作者	Zhang, Wensheng; Tian, Qi
作者单位	1.Huawei Technol, Cloud BU, Shenzhen 51800, Peoples R China 2.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing 101408, Peoples R China 3.Chinese Acad Sci, Inst Automat, Res Ctr Precis Sensing & Control, Beijing 100190, Peoples R China 4.East China Normal Univ, Sch Comp Sci & Software Engn, Shanghai 200241, Peoples R China
推荐引用方式 GB/T 7714	Zhang, Zhizhong,Xie, Yuan,Li, Ding,et al. Learning to Align via Wasserstein for Person Re-Identification[J]. IEEE TRANSACTIONS ON IMAGE PROCESSING,2020,29:7104-7116.
APA	Zhang, Zhizhong,Xie, Yuan,Li, Ding,Zhang, Wensheng,&Tian, Qi.(2020).Learning to Align via Wasserstein for Person Re-Identification.IEEE TRANSACTIONS ON IMAGE PROCESSING,29,7104-7116.
MLA	Zhang, Zhizhong,et al."Learning to Align via Wasserstein for Person Re-Identification".IEEE TRANSACTIONS ON IMAGE PROCESSING 29(2020):7104-7116.