Sensitivity-Adaptive Augmentation for Robust Segmentation
Keywords: augmentation, sensitivity analysis, robustness to corruption
TL;DR: We introduce a model-free augmentation policy based on sensitivity analysis to corruptions that is practical to online learning.
Abstract: Achieving robustness in image segmentation models is challenging due to the fine-grained nature of pixel-level classification. These models, which are crucial for many real-time perception applications, particularly struggle when faced with natural corruptions. While sensitivity analysis can help us understand how input variables influence model outputs, applying it to natural and uncontrollable corruptions in training data is difficult. In this work, we present an efficient, sensitivity-based augmentation method to enhance robustness against natural corruptions. Our sensitivity analysis approach runs up to 10 times faster and requires up to 200 times less storage than previous approaches, enabling practical, on-the-fly estimation during training for a model-free augmentation policy.
With minimal fine-tuning, our sensitivity-based augmentation method achieves improved robustness on both real-world and synthetic datasets compared to state-of-the-art data augmentation techniques in image segmentation tasks.
Supplementary Material: pdf
Primary Area: applications to computer vision, audio, language, and other modalities
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 3904
Loading