Delving into Decision-based Black-box Attacks on Semantic Segmentation
This addresses security risks in applications using semantic segmentation, such as autonomous driving, by demonstrating a novel and efficient attack method, though it is incremental in exploring a less-studied attack type.
The paper tackles the adversarial vulnerability of semantic segmentation models to black-box decision-based attacks, proposing the Discrete Linear Attack (DLA) that reduces PSPNet's mIoU from 77.83% to 2.14% with only 50 queries on Cityscapes.
Semantic segmentation is a fundamental visual task that finds extensive deployment in applications with security-sensitive considerations. Nonetheless, recent work illustrates the adversarial vulnerability of semantic segmentation models to white-box attacks. However, its adversarial robustness against black-box attacks has not been fully explored. In this paper, we present the first exploration of black-box decision-based attacks on semantic segmentation. First, we analyze the challenges that semantic segmentation brings to decision-based attacks through the case study. Then, to address these challenges, we first propose a decision-based attack on semantic segmentation, called Discrete Linear Attack (DLA). Based on random search and proxy index, we utilize the discrete linear noises for perturbation exploration and calibration to achieve efficient attack efficiency. We conduct adversarial robustness evaluation on 5 models from Cityscapes and ADE20K under 8 attacks. DLA shows its formidable power on Cityscapes by dramatically reducing PSPNet's mIoU from an impressive 77.83% to a mere 2.14% with just 50 queries.