Adversarial Bounding Boxes Generation (ABBG) Attack against Visual Object Trackers

TL;DR

This paper introduces a novel white-box adversarial attack method targeting transformer-based visual object trackers, generating adversarial bounding boxes to deceive trackers and outperform existing attacks on multiple datasets.

Contribution

A new attack approach for transformer trackers using only one bounding box to generate adversarial examples, expanding attack applicability.

Findings

Outperforms existing attacks on several transformer trackers.

Effective on multiple benchmark datasets.

Demonstrates robustness of the proposed attack method.

Abstract

Adversarial perturbations aim to deceive neural networks into predicting inaccurate results. For visual object trackers, adversarial attacks have been developed to generate perturbations by manipulating the outputs. However, transformer trackers predict a specific bounding box instead of an object candidate list, which limits the applicability of many existing attack scenarios. To address this issue, we present a novel white-box approach to attack visual object trackers with transformer backbones using only one bounding box. From the tracker predicted bounding box, we generate a list of adversarial bounding boxes and compute the adversarial loss for those bounding boxes. Experimental results demonstrate that our simple yet effective attack outperforms existing attacks against several robust transformer trackers, including TransT-M, ROMTrack, and MixFormer, on popular benchmark tracking datasets such as GOT-10k, UAV123, and VOT2022STS.