Poisoning the Pixels: Revisiting Backdoor Attacks on Semantic Segmentation

TL;DR

This paper systematically explores backdoor attacks on semantic segmentation models, introducing a unified framework called BADSEG, revealing vulnerabilities across architectures, and highlighting the ineffectiveness of current defenses.

Contribution

The paper formalizes new attack vectors for semantic segmentation, proposes BADSEG for optimized backdoor attacks, and demonstrates persistent vulnerabilities in recent models.

Findings

BADSEG achieves high attack success with minimal impact on clean data

Existing defenses fail to reliably mitigate the proposed backdoor attacks

Vulnerabilities are present in recent architectures including transformer-based models and SAM

Abstract

Semantic segmentation models are widely deployed in safety-critical applications such as autonomous driving, yet their vulnerability to backdoor attacks remains largely underexplored. Prior segmentation backdoor studies transfer threat settings from existing image classification tasks, focusing primarily on object-to-background mis-segmentation. In this work, we revisit the threats by systematically examining backdoor attacks tailored to semantic segmentation. We identify four coarse-grained attack vectors (Object-to-Object, Object-to-Background, Background-to-Object, and Background-to-Background attacks), as well as two fine-grained vectors (Instance-Level and Conditional attacks). To formalize these attacks, we introduce BADSEG, a unified framework that optimizes trigger designs and applies label manipulation strategies to maximize attack performance while preserving victim model utility. Extensive experiments across diverse segmentation architectures on benchmark datasets demonstrate that BADSEG achieves high attack effectiveness with minimal impact on clean samples. We further evaluate six representative defenses and find that they fail to reliably mitigate our attacks, revealing critical gaps in current defenses. Finally, we demonstrate that these vulnerabilities persist in recent emerging architectures, including transformer-based networks and the Segment Anything Model (SAM), thereby compromising their security. Our work reveals previously overlooked security vulnerabilities in semantic segmentation, and motivates the development of defenses tailored to segmentation-specific threat models.