OmniPatch: A Universal Adversarial Patch for ViT-CNN Cross-Architecture Transfer in Semantic Segmentation

TL;DR

OmniPatch is a universal adversarial patch designed to attack both ViT and CNN semantic segmentation models across different architectures without needing model details, enhancing robustness testing for autonomous driving systems.

Contribution

The paper introduces OmniPatch, a novel training framework for creating universal patches that transfer across architectures and models without access to target parameters.

Findings

OmniPatch successfully attacks multiple architectures.

The patch generalizes across different models.

It improves robustness testing for autonomous systems.

Abstract

Robust semantic segmentation is crucial for safe autonomous driving, yet deployed models remain vulnerable to black-box adversarial attacks when target weights are unknown. Most existing approaches either craft image-wide perturbations or optimize patches for a single architecture, which limits their practicality and transferability. We introduce OmniPatch, a training framework for learning a universal adversarial patch that generalizes across images and both ViT and CNN architectures without requiring access to target model parameters.