Learning Spatial Distribution of Long-Term Trackers Scores

TL;DR

This paper introduces a learning-based fusion approach for long-term trackers that generalizes to multiple baseline trackers, improving re-detection and achieving state-of-the-art recall on benchmark datasets.

Contribution

It proposes a novel method to learn the correlation of outcomes among multiple trackers, extending fusion strategies to an arbitrary number of trackers for long-term tracking.

Findings

Recall of 0.738 on LTB-50 dataset when trained on VOT-LT2022

Recall of 0.619 when reversing datasets, showing model robustness

Results are highly competitive with state-of-the-art methods

Abstract

Long-Term tracking is a hot topic in Computer Vision. In this context, competitive models are presented every year, showing a constant growth rate in performances, mainly measured in standardized protocols as Visual Object Tracking (VOT) and Object Tracking Benchmark (OTB). Fusion-trackers strategy has been applied over last few years for overcoming the known re-detection problem, turning out to be an important breakthrough. Following this approach, this work aims to generalize the fusion concept to an arbitrary number of trackers used as baseline trackers in the pipeline, leveraging a learning phase to better understand how outcomes correlate with each other, even when no target is present. A model and data independence conjecture will be evidenced in the manuscript, yielding a recall of 0.738 on LTB-50 dataset when learning from VOT-LT2022, and 0.619 by reversing the two datasets. In both cases, results are strongly competitive with state-of-the-art and recall turns out to be the first on the podium.