Title: A state-of-the-art review on person re-identification with deep learning
Authors: Peng Gao; Xiao Yue; Wei Chen; Weidong Fang; Zijian Tian; Fan Zhang
Addresses: Engineering Research Center of Mine Digitalization of Ministry of Education of Computer Science and Technology, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China ' Engineering Research Center of Mine Digitalization of Ministry of Education of Computer Science and Technology, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China ' Engineering Research Center of Mine Digitalization of Ministry of Education of Computer Science and Technology, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China ' Key Laboratory of Wireless Sensor Network & Communication, Shanghai Institute of Micro-system and Information Technology, Chinese Academy of Sciences, and Shanghai Research Center for Wireless Communications, Shanghai 201210, China ' School of Mechanical Electronic and Information Engineering, China University of Mining and Technology (Beijing), Beijing, 100083, China ' School of Mechanical Electronic and Information Engineering, China University of Mining and Technology (Beijing), Beijing, 100083, China
Abstract: Person re-identification (ReID), as a sub-direction of computer vision, has attracted more and more attention. In recent years, we have witnessed significant progress of person ReID driven by deep neural network architectures. In this paper, we introduce the progress of person ReID based on deep learning in recent years, including representation learning methods, metric learning methods, part-based methods, GAN-based methods, and video-based methods. The class of methods are summarised and analysed, and then we introduce the image-based datasets and the video-based datasets. We further discuss some of the current challenges and introduce some potential solutions in person ReID. Finally, we present the possible future directions of person ReID, such as collecting more abundant pedestrian datasets, adopting semi-supervised or unsupervised methods in person ReID. The purpose of this paper is to provide insights for the research on the person ReID and to present different methods of person ReID based on deep learning.
Keywords: person re-identification; computer vision; deep learning; review.
DOI: 10.1504/IJAHUC.2022.125425
International Journal of Ad Hoc and Ubiquitous Computing, 2022 Vol.41 No.2, pp.69 - 91
Received: 08 Jan 2021
Accepted: 01 Nov 2021
Published online: 09 Sep 2022 *