Person Re-Identification in Distributed Wide-Area Surveillance

Date

2014-05

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Person re-identification (Re-ID) is a fundamental task in automated video surveillance and has been an area of intense research in the past few years. Given an image or video of a person taken from one camera, re-identification is the process of identifying the person from images or videos taken from a different camera. Re-ID is indispensable in establishing consistent labeling across multiple cameras or even within the same camera to re-establish disconnected or lost tracks. Apart from surveillance it has applications in robotics, multimedia, and forensics. Person re-identification is a diffcult problem because of the visual ambiguity and spatio-temporal uncertainty in a person's appearance across different cameras. However, the problem has received significant attention from the computer-vision-research community due to its wide applicability and utility. In this work, we explore the problem of person re-identification for multi-camera tracking, to understand the nature of Re-ID, constraints and conditions under which it is to be addressed and possible solutions to each aspect. We show that Re-ID for multi-camera tracking is inherently an open set Re-ID problem with dynamically evolving gallery and open probe set. We propose multi-feature person models for both single and multi-shot Re-ID with a focus on incorporating unique features suitable for short as well as long period Re-ID. Finally, we adapt a novelty detection technique to address the problem of open set Re-ID. In conclusion we identify the open issues in Re-ID like, long-period Re-ID and scalability along with a discussion on potential directions for further research.

Description

Keywords

Person re-identification, Open-set Re-ID, Closed-set Re-ID, Short-period Re-ID, Long-period Re-ID, Novelty detection

Citation

Portions of this document appear in: Bedagkar-Gala, Apurva, and Shishir K. Shah. "Part-based spatio-temporal model for multi-person re-identification." Pattern Recognition Letters 33, no. 14 (2012): 1908-1915.