Novel Approach to Out-of-Distribution Segmentation Using Wasserstein Uncertainty

This research explores a novel method for identifying out-of-distribution data in image segmentation using Wasserstein-based evidential uncertainty. The approach likely addresses a critical challenge in deploying segmentation models in real-world scenarios where unexpected data is encountered.

• •Proposes a new methodology for identifying out-of-distribution (OOD) samples in image segmentation.
• •Utilizes Wasserstein distance and evidential uncertainty for improved OOD detection.
• •Addresses the practical problem of robustness in real-world segmentation applications.