Multiple Object Tracking: A Literature Review

Multiple Object Tracking is an important computer vision task which has gained increasing attention due to its academic and commercial potential. Although different approaches have been proposed to tackle it, there still exist many issues unsolved. In order to help readers understand this topic, we contribute a systematic and comprehensive review. In the review, we inspect recent advances in various aspects and propose some interesting directions for future research. To our best knowledge, there has not been any review about this topic in the community. We endeavor to provide a thorough review on the development of this problem in the last decades. The main contributions are fourfold: 1) Key aspects in a multiple object tracking system, including how to formulate MOT generally, how to categorize MOT algorithms, what needs to be considered when developing a MOT system and how to evaluate a MOT system, are discussed from the perspective of understanding a topic. We believe this could not only provide researchers, especially new comers to the topic of MOT, a general understanding of the state of the arts, but also help them to comprehend the essential components of a MOT system and the inter-component connection. 2) Instead of enumerating individual works, we discuss existing work according to the various aspects involved in a MOT system. In each aspect, methods are divided into different groups and each group is discussed in details for the principles, advances and drawbacks. 3) We examine experiments of existing publications and give tables which list results on the popular data sets to provide convenient comparison. We also provide some interesting discoveries by analyzing these tables. 4) We offer some potential directions and respective discussions about MOT, which are still open issues and need more research efforts. This would be helpful to identify interesting problems.