From Pixel to Mask: A Survey of Out-of-Distribution Segmentation
It addresses the need for precise OoD segmentation in autonomous driving to enhance system robustness, but it is incremental as a survey paper.
This survey tackles the problem of out-of-distribution (OoD) segmentation, which localizes anomalous objects at pixel-level granularity for safety-critical applications like autonomous driving, by reviewing recent advances, grouping approaches into four categories, and discussing challenges and future directions.
Out-of-distribution (OoD) detection and segmentation have attracted growing attention as concerns about AI security rise. Conventional OoD detection methods identify the existence of OoD objects but lack spatial localization, limiting their usefulness in downstream tasks. OoD segmentation addresses this limitation by localizing anomalous objects at pixel-level granularity. This capability is crucial for safety-critical applications such as autonomous driving, where perception modules must not only detect but also precisely segment OoD objects, enabling targeted control actions and enhancing overall system robustness. In this survey, we group current OoD segmentation approaches into four categories: (i) test-time OoD segmentation, (ii) outlier exposure for supervised training, (iii) reconstruction-based methods, (iv) and approaches that leverage powerful models. We systematically review recent advances in OoD segmentation for autonomous-driving scenarios, identify emerging challenges, and discuss promising future research directions.