Uncertainty-Based Detection of Adversarial Attacks in Semantic Segmentation

Kira Maag, Asja Fischer

2024

Abstract

State-of-the-Art deep neural networks have proven to be highly powerful in a broad range of tasks, including semantic image segmentation. However, these networks are vulnerable against adversarial attacks, i.e., non-perceptible perturbations added to the input image causing incorrect predictions, which is hazardous in safety-critical applications like automated driving. Adversarial examples and defense strategies are well studied for the image classification task, while there has been limited research in the context of semantic segmentation. First works however show that the segmentation outcome can be severely distorted by adversarial attacks. In this work, we introduce an uncertainty-based approach for the detection of adversarial attacks in semantic segmentation. We observe that uncertainty as for example captured by the entropy of the output distribution behaves differently on clean and perturbed images and leverage this property to distinguish between the two cases. Our method works in a light-weight and post-processing manner, i.e., we do not modify the model or need knowledge of the process used for generating adversarial examples. In a thorough empirical analysis, we demonstrate the ability of our approach to detect perturbed images across multiple types of adversarial attacks.

Paper Citation

in Harvard Style

Maag K. and Fischer A. (2024). Uncertainty-Based Detection of Adversarial Attacks in Semantic Segmentation. In Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP; ISBN 978-989-758-679-8, SciTePress, pages 37-46. DOI: 10.5220/0012303500003660

in Bibtex Style

@conference{visapp24,
author={Kira Maag and Asja Fischer},
title={Uncertainty-Based Detection of Adversarial Attacks in Semantic Segmentation},
booktitle={Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP},
year={2024},
pages={37-46},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012303500003660},
isbn={978-989-758-679-8},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP
TI - Uncertainty-Based Detection of Adversarial Attacks in Semantic Segmentation
SN - 978-989-758-679-8
AU - Maag K.
AU - Fischer A.
PY - 2024
SP - 37
EP - 46
DO - 10.5220/0012303500003660
PB - SciTePress