AAformer: Auto-Aligned Transformer for Person Re-Identification

doi:10.1109/TNNLS.2023.3301856

CORC > 自动化研究所 > 中国科学院自动化研究所 > 紫东太初大模型研究中心

	AAformer: Auto-Aligned Transformer for Person Re-Identification
	Zhu, Kuan 4; Guo, Haiyun 3,4,6; Zhang, Shiliang 7; Wang, Yaowei 2; Liu, Jing 4,6; Wang, Jinqiao 1,4,5,6; Tang, Ming 4
刊名	IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
	2023-08-25
页码	11
关键词	Auto-alignment part-level representation person re-identification (re-ID) transformer
ISSN号	2162-237X
DOI	10.1109/TNNLS.2023.3301856
通讯作者	Guo, Haiyun(haiyun.guo@nlpr.ia.ac.cn)
英文摘要	In person re-identification (re-ID), extracting part-level features from person images has been verified to be crucial to offer fine-grained information. Most of the existing CNN-based methods only locate the human parts coarsely, or rely on pretrained human parsing models and fail in locating the identifiable nonhuman parts (e.g., knapsack). In this article, we introduce an alignment scheme in transformer architecture for the first time and propose the auto-aligned transformer (AAformer) to automatically locate both the human parts and nonhuman ones at patch level. We introduce the "Part tokens (PARTs)", which are learnable vectors, to extract part features in the transformer. A PART only interacts with a local subset of patches in self-attention and learns to be the part representation. To adaptively group the image patches into different subsets, we design the auto-alignment. Auto-alignment employs a fast variant of optimal transport (OT) algorithm to online cluster the patch embeddings into several groups with the PARTs as their prototypes. AAformer integrates the part alignment into the self-attention and the output PARTs can be directly used as part features for retrieval. Extensive experiments validate the effectiveness of PARTs and the superiority of AAformer over various state-of-the-art methods.
资助项目	Key-Area Research and Development Program of Guangdong Province[2021B0101410003] ; National Natural Science Foundation of China[62276260] ; National Natural Science Foundation of China[62002356] ; National Natural Science Foundation of China[61976210]
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:001060547200001
资助机构	Key-Area Research and Development Program of Guangdong Province ; National Natural Science Foundation of China
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/53197]
专题	紫东太初大模型研究中心
通讯作者	Guo, Haiyun
作者单位	1.Wuhan AI Res, Wuhan 430073, Peoples R China 2.Peng Cheng Lab, Shenzhen 518066, Peoples R China 3.Dev Res Inst Guangzhou Smart City, Guangzhou 510805, Peoples R China 4.Chinese Acad Sci, Inst Automat, Fdn Model Res Ctr, Beijing 100190, Peoples R China 5.Peng Cheng Lab, Shenzhen 518066, Peoples R China 6.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing 100049, Peoples R China 7.Peking Univ, Sch Comp Sci, Natl Key Lab Multimedia Informat Proc, Beijing 100871, Peoples R China
推荐引用方式 GB/T 7714	Zhu, Kuan,Guo, Haiyun,Zhang, Shiliang,et al. AAformer: Auto-Aligned Transformer for Person Re-Identification[J]. IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS,2023:11.
APA	Zhu, Kuan.,Guo, Haiyun.,Zhang, Shiliang.,Wang, Yaowei.,Liu, Jing.,...&Tang, Ming.(2023).AAformer: Auto-Aligned Transformer for Person Re-Identification.IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS,11.
MLA	Zhu, Kuan,et al."AAformer: Auto-Aligned Transformer for Person Re-Identification".IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS (2023):11.