RPT++: Customized Feature Representation for Siamese Visual Tracking

TL;DR

This paper introduces RPT++, a visual tracking method that uses task-specific feature extractors to improve classification and boundary estimation, achieving state-of-the-art results on multiple benchmarks.

Contribution

It proposes two novel feature extractors, polar pooling and extreme pooling, to address feature misalignment in visual tracking.

Findings

RPT++ outperforms existing trackers on multiple benchmarks.

Task-specific features improve classification and boundary accuracy.

The method achieves new state-of-the-art performance on OTB-100, VOT, GOT-10k, TrackingNet, and LaSOT.

Abstract

While recent years have witnessed remarkable progress in the feature representation of visual tracking, the problem of feature misalignment between the classification and regression tasks is largely overlooked. The approaches of feature extraction make no difference for these two tasks in most of advanced trackers. We argue that the performance gain of visual tracking is limited since features extracted from the salient area provide more recognizable visual patterns for classification, while these around the boundaries contribute to accurately estimating the target state. We address this problem by proposing two customized feature extractors, named polar pooling and extreme pooling to capture task-specific visual patterns. Polar pooling plays the role of enriching information collected from the semantic keypoints for stronger classification, while extreme pooling facilitates explicit visual patterns of the object boundary for accurate target state estimation. We demonstrate the effectiveness of the task-specific feature representation by integrating it into the recent and advanced tracker RPT. Extensive experiments on several benchmarks show that our Customized Features based RPT (RPT++) achieves new state-of-the-art performances on OTB-100, VOT2018, VOT2019, GOT-10k, TrackingNet and LaSOT.