Revisiting the details when evaluating a visual tracker

TL;DR

This paper revisits the evaluation process of visual trackers, proposing a simpler and more accurate method, and highlights the importance of detailed analysis over seeking a single best tracker.

Contribution

Introduces a new evaluation method for visual trackers that is simpler, more accurate, and extensible, improving comparison and analysis of tracking algorithms.

Findings

No absolute best tracker among current algorithms

Detailed analysis is essential for selecting suitable trackers

Proposed method enhances tracker evaluation accuracy

Abstract

Visual tracking algorithms are naturally adopted in various applications, there have been several benchmarks and many tracking algorithms, more expected to appear in the future. In this report, I focus on single object tracking and revisit the details of tracker evaluation based on widely used OTB\cite{otb} benchmark by introducing a simpler, accurate, and extensible method for tracker evaluation and comparison. Experimental results suggest that there may not be an absolute winner among tracking algorithms. We have to perform detailed analysis to select suitable trackers for use cases.