Erosion Attack for Adversarial Training to Enhance Semantic Segmentation Robustness

2026-01-21

AI Summary

This work addresses the insufficient robustness of existing semantic segmentation models under adversarial attacks, a limitation exacerbated by conventional adversarial training methods that overlook internal contextual semantic relationships within samples. To overcome this, the authors propose EroSeg-AT, a novel framework that introduces an erosion-based perturbation propagation mechanism grounded in pixel sensitivity and semantic consistency. Specifically, sensitive regions are first identified based on pixel-wise confidence scores; perturbations are then progressively propagated from low-confidence to high-confidence regions to disrupt semantic coherence and generate more potent adversarial examples. Experimental results demonstrate that this approach significantly enhances the effectiveness of adversarial attacks and substantially improves the robustness of semantic segmentation models when integrated into adversarial training.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Search and Optimization: Adversarial Search

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Existing segmentation models exhibit significant vulnerability to adversarial attacks.To improve robustness, adversarial training incorporates adversarial examples into model training. However, existing attack methods consider only global semantic information and ignore contextual semantic relationships within the samples, limiting the effectiveness of adversarial training. To address this issue, we propose EroSeg-AT, a vulnerability-aware adversarial training framework that leverages EroSeg to generate adversarial examples. EroSeg first selects sensitive pixels based on pixel-level confidence and then progressively propagates perturbations to higher-confidence pixels, effectively disrupting the semantic consistency of the samples. Experimental results show that, compared to existing methods, our approach significantly improves attack effectiveness and enhances model robustness under adversarial training.

Problem

• adversarial training
• semantic segmentation
• adversarial attack
• semantic consistency
• robustness

Innovation

• adversarial training
• semantic segmentation
• erosion attack
• semantic consistency
• pixel-level perturbation