Person re-identification: A taxonomic survey and the path ahead

Abstract

Person re-identification (PRId) is one of the most challenging tasks in automated video surveillance and has been an area of intense research spanning the past decade. PRId aims at finding a person who has previously been identified using some unique descriptor of the person. This survey comprises a wide spectrum of PRId methods spanning from traditional to deep learning-based being analyzed and compared. This survey also discusses different PRId frameworks on the basis of machine learning and deep learning. It offers a multi-dimensional taxonomy to classify the most pertinent researches according to different perspectives and tries to unify the categorization of PRId methods and fill the gap between the recently published surveys. This study highlights the challenges in building PRId systems. It presents a critical overview of recent progress and the state-of-the-art approaches to solving some major challenges of existing PRId systems. Furthermore, we discuss the performance comparisons of the various state-of-the-art in different datasets. Finally, we discuss several open issues and directions for future studies.