ECCV 2022 Workshop on Adversarial Robustness in the Real World

Overview

Recent deep-learning-based methods achieve great performance on various vision applications. However, insufficient robustness on adversarial cases limits real-world applications of deep-learning-based methods. AROW workshop aims to explore adversarial examples, as well as, evaluate and improve the adversarial robustness of computer vision systems.

Topics of AROW workshop include but are not limited to:

Improving model robustness against unrestricted adversarial attacks

Improving generalization to out-of-distribution samples or unforeseen adversaries

Discovery of real-world adversarial examples

Novel architectures with robustness to occlusion, viewpoint, and other real-world domain shifts

Domain adaptation techniques for robust vision in the real world

Datasets for evaluating model robustness

Structured deep models and explainable AI

Prize

The workshop is sponsored by the Future Fund regranting program. The funding covers three Best Paper Awards ($10,000 each, $30,000 in total). The awarded papers should study model robustness to threats beyond small l_p perturbations (e.g., adversarially optimized fog and snow effects, adversarial patches, adversarial elastic distortions, new attacks, etc.). The best papers will research attacks with large budgets that are perceptible, and/or attacks with specifications that are not known beforehand and are unforeseen.

Submission

Submission format:

Submissions need to be anonymized and follow the ECCV 2022 Author Instructions. The workshop considers two types of submissions: (1) Long Paper: Papers are limited to 14 pages excluding references and will be included in the official ECCV proceedings, Please use the ECCV template ; (2) Short Papers: Papers are limited to 4 pages including references and will NOT be included in the official ECCV proceedings (does not count as double submission for most vision conferences). Please use the CVPR template for the short papers.