Structured Universal Adversarial Attacks on Object Detection for Video Sequences
Abstract
Video-based object detection plays a vital role in safety-critical applications. While deep learning-based object detectors have achieved impressive performance, they remain vulnerable to adversarial attacks, particularly those involving universal perturbations. In this work, we propose a minimally distorted universal adversarial attack tailored for video object detection, which leverages nuclear norm regularization to promote structured perturbations concentrated in the background. To optimize this formulation efficiently, we employ an adaptive, optimistic exponentiated gradient method that enhances both scalability and convergence. Our results demonstrate that the proposed attack outperforms both low-rank projected gradient descent and Frank-Wolfe based attacks in effectiveness while maintaining high stealthiness. All code and data are publicly available at https://github.com/jsve96/AO-Exp-Attack.
Cite
@article{arxiv.2510.14460,
title = {Structured Universal Adversarial Attacks on Object Detection for Video Sequences},
author = {Sven Jacob and Weijia Shao and Gjergji Kasneci},
journal= {arXiv preprint arXiv:2510.14460},
year = {2025}
}
Comments
Accepted at GCPR 2025 (German Conference on Pattern Recognition). This is a different version as submitted to the conference, not the official conference proceedings