Multi-Object Tracking with Siamese Track-RCNN

TL;DR

This paper introduces Siamese Track-RCNN, a unified multi-object tracking system that combines detection, motion estimation, and re-identification in a single framework, achieving superior accuracy and efficiency on MOTChallenge datasets.

Contribution

The paper presents a novel unified tracking architecture that integrates detection, motion estimation, and re-identification, simplifying the multi-object tracking pipeline.

Findings

Outperforms state-of-the-art on MOTChallenge datasets

Achieves higher accuracy with increased efficiency

Successfully unifies multiple tracking components into one system

Abstract

Multi-object tracking systems often consist of a combination of a detector, a short term linker, a re-identification feature extractor and a solver that takes the output from these separate components and makes a final prediction. Differently, this work aims to unify all these in a single tracking system. Towards this, we propose Siamese Track-RCNN, a two stage detect-and-track framework which consists of three functional branches: (1) the detection branch localizes object instances; (2) the Siamese-based track branch estimates the object motion and (3) the object re-identification branch re-activates the previously terminated tracks when they re-emerge. We test our tracking system on two popular datasets of the MOTChallenge. Siamese Track-RCNN achieves significantly higher results than the state-of-the-art, while also being much more efficient, thanks to its unified design.