Visual Object Tracking: The Initialisation Problem

TL;DR

This paper addresses the initialisation challenge in visual object tracking by framing it as a missing labels problem and evaluating three segmentation techniques, highlighting LBDM's robustness and improved performance.

Contribution

It introduces a novel approach to initialisation in object tracking using label learning techniques and compares their effectiveness on a standard benchmark.

Findings

LBDM outperforms OC-SVM and SBBM with cross-validated parameters.

OC-SVM and SBBM are accurate but highly parameter-dependent.

LBDM is robust to parameter variation and suitable for real-world use.

Abstract

Model initialisation is an important component of object tracking. Tracking algorithms are generally provided with the first frame of a sequence and a bounding box (BB) indicating the location of the object. This BB may contain a large number of background pixels in addition to the object and can lead to parts-based tracking algorithms initialising their object models in background regions of the BB. In this paper, we tackle this as a missing labels problem, marking pixels sufficiently away from the BB as belonging to the background and learning the labels of the unknown pixels. Three techniques, One-Class SVM (OC-SVM), Sampled-Based Background Model (SBBM) (a novel background model based on pixel samples), and Learning Based Digital Matting (LBDM), are adapted to the problem. These are evaluated with leave-one-video-out cross-validation on the VOT2016 tracking benchmark. Our evaluation shows both OC-SVMs and SBBM are capable of providing a good level of segmentation accuracy but are too parameter-dependent to be used in real-world scenarios. We show that LBDM achieves significantly increased performance with parameters selected by cross validation and we show that it is robust to parameter variation.