Trends in Vehicle Re-Identification Past, Present, and Future: A Comprehensive Review

Abstract

Vehicle Re-identification (re-id) over surveillance camera network with non-overlapping field of view is an exciting and challenging task in intelligent transportation systems (ITS). Due to its versatile applicability in metropolitan cities, it gained significant attention. Vehicle re-id matches targeted vehicle over non-overlapping views in multiple camera network. However, it becomes more difficult due to inter-class similarity, intra-class variability, viewpoint changes, and spatio-temporal uncertainty. In order to draw a detailed picture of vehicle re-id research, this paper gives a comprehensive description of the various vehicle re-id technologies, applicability, datasets, and a brief comparison of different methodologies. Our paper specifically focuses on vision-based vehicle re-id approaches, including vehicle appearance, license plate, and spatio-temporal characteristics. In addition, we explore the main challenges as well as a variety of applications in different domains. Lastly, a detailed comparison of current state-of-the-art methods performances over VeRi-776 and VehicleID datasets is summarized with future directions. We aim to facilitate future research by reviewing the work being done on vehicle re-id till to date.

FRS-Net: An Efficient Ship Detection Network for Thin-Cloud and Fog-Covered High-Resolution Optical Satellite Imagery

“…These coefficients are used in the above model to add fog to the original dataset image and quantitatively control the fog occlusion thickness through specific coefficients during the fogging process. The specific process is as follows: t(x)11-ω⋅min c∈(r,g,b) (min y∈Ω(x) I c (y) A c ) (7) where ω is set to 0.95. Ω(x) represents the local area block centered on "x".…”

“…HIP detection technology based on optical remote sensing images is widely used in river monitoring [1][2][3][4], port management [5][6][7][8] and illegal border crossing detection [9][10]. Since ships are typical moving targets, such tasks have high requirements on timeliness, and the traditional satel-Zhiqi Zhang, Huigang Zheng, Jinshan Cao and Xiaoxiao Feng are with School of Computer Science, Hubei University of Technology, Wuhan 430068, China (e-mail: zzq540@hbut.edu.cn; 102011015@hbut.edu.cn; caojs@hbut.edu.cn; 20220026@hbut.edu.cn ) lite remote sensing application mode that focuses on post-processing of high-quality images is difficult to meet the needs.…”

SBNet: Segmentation-based Network for Natural Language-based Vehicle Search

“…The total loss function is defined as follows: (23) where L seg vehicle segmentation loss, L cls is vehicle classification loss, L sub is substitution loss, L f ut is future prediction loss, and λ 1 and λ 2 are weights and are set to 0.2 and 0.2 respectively. The losses are described in the following sections.…”

“…Vehicle retrieval using image-based queries is called vehicle re-identification in computer vision [11,23,13,14]. Given a query image, the algorithm obtains affinities be- tween the query and the vehicle images in the database and retrieves the most similar vehicles.…”

Enhanced Detection Model and Joint Scoring Strategy for Multi-Vehicle Tracking

“…With the noticeable progress of deep learning, a huge variety of deep neural network-based trackers have been proposed for MOT tasks [7]. Among them, one-shot MOT methods, which simultaneously accomplish target detection and identity embedding re-identification [8], have begun to gain significant attention. Considering that one-short MOT approaches have the advantages of high reliability and low computation cost, they are very suitable for multi-vehicle tracking in practical traffic scenarios.…”