Tracking-by-Trackers with a Distilled and Reinforced Model

TL;DR

This paper introduces a unified tracking approach using a distilled and reinforced model that leverages multiple trackers to create fast, adaptable, and fusion-capable visual object trackers, validated by extensive experiments.

Contribution

A novel methodology combining knowledge distillation and reinforcement learning to unify and enhance visual tracking capabilities.

Findings

Achieves real-time performance comparable to state-of-the-art trackers

Enables fast single-shot tracking with online adaptation

Supports effective fusion of multiple trackers

Abstract

Visual object tracking was generally tackled by reasoning independently on fast processing algorithms, accurate online adaptation methods, and fusion of trackers. In this paper, we unify such goals by proposing a novel tracking methodology that takes advantage of other visual trackers, offline and online. A compact student model is trained via the marriage of knowledge distillation and reinforcement learning. The first allows to transfer and compress tracking knowledge of other trackers. The second enables the learning of evaluation measures which are then exploited online. After learning, the student can be ultimately used to build (i) a very fast single-shot tracker, (ii) a tracker with a simple and effective online adaptation mechanism, (iii) a tracker that performs fusion of other trackers. Extensive validation shows that the proposed algorithms compete with real-time state-of-the-art trackers.