Efficient universal shuffle attack for visual object tracking

TL;DR

This paper introduces an offline universal adversarial attack method that efficiently disrupts deep visual object trackers across multiple videos using a single perturbation, addressing real-time and re-initialization challenges.

Contribution

The paper proposes the Efficient Universal Shuffle Attack, a novel offline universal attack method that improves attack efficiency and effectiveness against state-of-the-art trackers.

Findings

EUSA significantly degrades tracker performance on OTB2015 and VOT2018.

The greedy gradient strategy enhances attack efficiency.

Triple loss improves attack success rate.

Abstract

Recently, adversarial attacks have been applied in visual object tracking to deceive deep trackers by injecting imperceptible perturbations into video frames. However, previous work only generates the video-specific perturbations, which restricts its application scenarios. In addition, existing attacks are difficult to implement in reality due to the real-time of tracking and the re-initialization mechanism. To address these issues, we propose an offline universal adversarial attack called Efficient Universal Shuffle Attack. It takes only one perturbation to cause the tracker malfunction on all videos. To improve the computational efficiency and attack performance, we propose a greedy gradient strategy and a triple loss to efficiently capture and attack model-specific feature representations through the gradients. Experimental results show that EUSA can significantly reduce the performance of state-of-the-art trackers on OTB2015 and VOT2018.