Staple: Complementary Learners for Real-Time Tracking

TLDR

Correlation filter trackers excel in robustness to motion blur and illumination changes but are sensitive to deformation, whereas color‑based models handle shape variation yet struggle with inconsistent lighting and limited discriminative power. The authors aim to show that a lightweight ridge‑regression tracker that fuses complementary cues can run at over 80 FPS while surpassing VOT14 entries and recent sophisticated trackers. The resulting tracker achieves >80 FPS and outperforms VOT14 entries and recent sophisticated trackers across multiple benchmarks.

Abstract

Correlation Filter-based trackers have recently achieved excellent performance, showing great robustness to challenging situations exhibiting motion blur and illumination changes. However, since the model that they learn depends strongly on the spatial layout of the tracked object, they are notoriously sensitive to deformation. Models based on colour statistics have complementary traits: they cope well with variation in shape, but suffer when illumination is not consistent throughout a sequence. Moreover, colour distributions alone can be insufficiently discriminative. In this paper, we show that a simple tracker combining complementary cues in a ridge regression framework can operate faster than 80 FPS and outperform not only all entries in the popular VOT14 competition, but also recent and far more sophisticated trackers according to multiple benchmarks.