Seeing Isn't Believing: Context-Aware Adversarial Patch Synthesis via Conditional GAN

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kazoom, Roie, Goldberg, Alon, Cohen, Hodaya, Hadar, Ofer
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911340339134464
author Kazoom, Roie
Goldberg, Alon
Cohen, Hodaya
Hadar, Ofer
author_facet Kazoom, Roie
Goldberg, Alon
Cohen, Hodaya
Hadar, Ofer
contents Adversarial patch attacks pose a severe threat to deep neural networks, yet most existing approaches rely on unrealistic white-box assumptions, untargeted objectives, or produce visually conspicuous patches that limit real-world applicability. In this work, we introduce a novel framework for fully controllable adversarial patch generation, where the attacker can freely choose both the input image x and the target class y target, thereby dictating the exact misclassification outcome. Our method combines a generative U-Net design with Grad-CAM-guided patch placement, enabling semantic-aware localization that maximizes attack effectiveness while preserving visual realism. Extensive experiments across convolutional networks (DenseNet-121, ResNet-50) and vision transformers (ViT-B/16, Swin-B/16, among others) demonstrate that our approach achieves state-of-the-art performance across all settings, with attack success rates (ASR) and target-class success (TCS) consistently exceeding 99%. Importantly, we show that our method not only outperforms prior white-box attacks and untargeted baselines, but also surpasses existing non-realistic approaches that produce detectable artifacts. By simultaneously ensuring realism, targeted control, and black-box applicability-the three most challenging dimensions of patch-based attacks-our framework establishes a new benchmark for adversarial robustness research, bridging the gap between theoretical attack strength and practical stealthiness.
format Preprint
id arxiv_https___arxiv_org_abs_2509_22836
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Seeing Isn't Believing: Context-Aware Adversarial Patch Synthesis via Conditional GAN
Kazoom, Roie
Goldberg, Alon
Cohen, Hodaya
Hadar, Ofer
Computer Vision and Pattern Recognition
Artificial Intelligence
Adversarial patch attacks pose a severe threat to deep neural networks, yet most existing approaches rely on unrealistic white-box assumptions, untargeted objectives, or produce visually conspicuous patches that limit real-world applicability. In this work, we introduce a novel framework for fully controllable adversarial patch generation, where the attacker can freely choose both the input image x and the target class y target, thereby dictating the exact misclassification outcome. Our method combines a generative U-Net design with Grad-CAM-guided patch placement, enabling semantic-aware localization that maximizes attack effectiveness while preserving visual realism. Extensive experiments across convolutional networks (DenseNet-121, ResNet-50) and vision transformers (ViT-B/16, Swin-B/16, among others) demonstrate that our approach achieves state-of-the-art performance across all settings, with attack success rates (ASR) and target-class success (TCS) consistently exceeding 99%. Importantly, we show that our method not only outperforms prior white-box attacks and untargeted baselines, but also surpasses existing non-realistic approaches that produce detectable artifacts. By simultaneously ensuring realism, targeted control, and black-box applicability-the three most challenging dimensions of patch-based attacks-our framework establishes a new benchmark for adversarial robustness research, bridging the gap between theoretical attack strength and practical stealthiness.
title Seeing Isn't Believing: Context-Aware Adversarial Patch Synthesis via Conditional GAN
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2509.22836