Abstract:Addressing the vulnerability of deep neural networks (DNNs) has attracted significant attention in recent years. While recent studies on adversarial attack and defense mainly reside in a single image, few efforts have been made to perform temporal attacks against video sequences. As the temporal consistency between frames is not considered, existing adversarial attack approaches designed for static images do not perform well for deep object tracking. In this work, we generate adversarial examples on top of video sequences to improve the tracking robustness against adversarial attacks under white-box and black-box settings. To this end, we consider motion signals when generating lightweight perturbations over the estimated tracking results frame-by-frame. For the white-box attack, we generate temporal perturbations via known trackers to degrade significantly the tracking performance. We transfer the generated perturbations into unknown targeted trackers for the black-box attack to achieve transferring attacks. Furthermore, we train universal adversarial perturbations and directly add them into all frames of videos, improving the attack effectiveness with minor computational costs. On the other hand, we sequentially learn to estimate and remove the perturbations from input sequences to restore the tracking performance. We apply the proposed adversarial attack and defense approaches to state-of-the-art tracking algorithms. Extensive evaluations on large-scale benchmark datasets, including OTB, VOT, UAV123, and LaSOT, demonstrate that our attack method degrades the tracking performance significantly with favorable transferability to other backbones and trackers. Notably, the proposed defense method restores the original tracking performance to some extent and achieves additional performance gains when not under adversarial attacks.

Unsupervised cycle-consistent adversarial attacks for visual object tracking

F&F Attack: Adversarial Attack Against Multiple Object Trackers by Inducing False Negatives and False Positives

One-Shot Adversarial Attacks on Visual Tracking With Dual Attention

Pluggable Attack for Visual Object Tracking

Efficient Adversarial Attacks for Visual Object Tracking

LRR: Language-Driven Resamplable Continuous Representation against Adversarial Tracking Attacks

IoU Attack: Towards Temporally Coherent Black-Box Adversarial Attack for Visual Object Tracking

Context-Guided Black-Box Attack for Visual Tracking

Hijacking Tracker: A Powerful Adversarial Attack on Visual Tracking.

A Unified Multi-Scenario Attacking Network for Visual Object Tracking.

Ad2Attack: Adaptive Adversarial Attack on Real-Time UAV Tracking

Universal Adversarial Attack Against 3D Object Tracking.

Towards Universal Physical Attacks on Single Object Tracking

Non-Rigid Transformation Based Adversarial Attack Against 3d Object Tracking

Robust Deep Object Tracking against Adversarial Attacks

Fooling Detection Alone is Not Enough: First Adversarial Attack against Multiple Object Tracking

Adversarial Examples in Visual Object Tracking in Satellite Videos: Cross-Frame Momentum Accumulation for Adversarial Examples Generation

Topology-aware Universal Adversarial Attack on 3D Object Tracking

SPARK: Spatial-Aware Online Incremental Attack Against Visual Tracking

Only Once Attack: Fooling the Tracker with Adversarial Template

STA: Adversarial Attacks on Siamese Trackers