🤖 AI Summary
This work addresses the challenge of detecting road anomalies—such as unknown obstacles—in open-world autonomous driving, where semantic segmentation models often fail to recognize out-of-distribution objects. The authors propose a lightweight, post-processing framework that requires neither model retraining nor access to anomalous data. Their approach innovatively integrates query-guided polygonal road geometry priors with zero-shot CLIP-based semantic filtering, supporting both in-distribution and generalized out-of-distribution prompts. By leveraging confidence analysis from a Mask Transformer, the method enables unsupervised and interpretable precise segmentation of anomalous regions. Evaluated on Fishyscapes, SMIYC, and RoadAnomaly benchmarks, the framework significantly outperforms existing methods, achieving state-of-the-art average precision on the Fishyscapes LostAndFound subset and demonstrating strong deployment feasibility and robustness.
📝 Abstract
Traditional semantic segmentation models operate under a closed-set assumption and struggle to recognize unknown or unexpected objects-an essential capability for autonomous driving. As a result, such models often misclassify or overlook out-of-distribution (OOD) road anomalies, posing safety risks in open-world environments. We present a lightweight, postprocessing, road-aware anomaly segmentation framework that requires no retraining, no OOD data, and no auxiliary supervision. Our approach builds on a mask transformer-based segmentation network by exploiting query-level mask confidence and deriving a polygonal road prior to detect gap regions that may correspond to anomalies. To further suppress false positives, we introduce a CLIP-based zero-shot semantic filtering module using in-distribution prompts, with optional generalized OOD prompts. By jointly leveraging spatial priors and semantic verification, our framework produces robust and interpretable anomaly predictions. Evaluation on three public benchmarks-Fishyscapes, SMIYC, and RoadAnomaly-shows consistently strong performance. In particular, our method outperforms the training-free baseline Maskomaly on most metrics and achieves the highest AP on Fishyscapes LostAndFound. These results demonstrate the practicality and deployability of our approach for real-world autonomous driving systems.