An Exploration of Target-Conditioned Segmentation Methods for Visual Object Trackers

TL;DR

This paper investigates how existing target-conditioned segmentation methods can be used to convert bounding-box trackers into segmentation trackers, enabling more precise object localization while maintaining real-time performance.

Contribution

It provides an extensive analysis of segmentation methods for transforming bounding-box trackers into segmentation trackers, demonstrating competitive performance with recent segmentation trackers.

Findings

Segmentation methods enable real-time target tracking.

Converted trackers compete with dedicated segmentation trackers.

The approach enhances object localization precision.

Abstract

Visual object tracking is the problem of predicting a target object's state in a video. Generally, bounding-boxes have been used to represent states, and a surge of effort has been spent by the community to produce efficient causal algorithms capable of locating targets with such representations. As the field is moving towards binary segmentation masks to define objects more precisely, in this paper we propose to extensively explore target-conditioned segmentation methods available in the computer vision community, in order to transform any bounding-box tracker into a segmentation tracker. Our analysis shows that such methods allow trackers to compete with recently proposed segmentation trackers, while performing quasi real-time.