Triggering Failures: Out-Of-Distribution detection by learning from local adversarial attacks in Semantic Segmentation

TL;DR

This paper introduces ObsNet, a novel out-of-distribution detection method for semantic segmentation that balances speed and accuracy by learning from local adversarial attacks and focusing on localized OOD regions.

Contribution

The paper presents ObsNet, a new OOD detection architecture with a training scheme based on local adversarial attacks, improving speed and accuracy in semantic segmentation.

Findings

Outperforms ten recent methods in speed and accuracy

Effective across multiple datasets

Validated through extensive ablation studies

Abstract

In this paper, we tackle the detection of out-of-distribution (OOD) objects in semantic segmentation. By analyzing the literature, we found that current methods are either accurate or fast but not both which limits their usability in real world applications. To get the best of both aspects, we propose to mitigate the common shortcomings by following four design principles: decoupling the OOD detection from the segmentation task, observing the entire segmentation network instead of just its output, generating training data for the OOD detector by leveraging blind spots in the segmentation network and focusing the generated data on localized regions in the image to simulate OOD objects. Our main contribution is a new OOD detection architecture called ObsNet associated with a dedicated training scheme based on Local Adversarial Attacks (LAA). We validate the soundness of our approach across numerous ablation studies. We also show it obtains top performances both in speed and accuracy when compared to ten recent methods of the literature on three different datasets.