A Dataset for Semantic Segmentation in the Presence of Unknowns

RA Cityscapes SMIYC-RA21 SMIYC-RO21 Standard Anomaly Evaluation Setting Proposed ISSU Benchmark Controlled evaluation Normal + low light 19 Classes + anomaly label Open-set evaluation ISSU -Train ISSU -Test -Static ISSU -Test -Temporal in-domain c r o s sd o m a i n cross-sensor, temporal Not controlled evaluation No evaluation of the primary objective Limited labels Mostly toy-like scenes Figure 1. Standard benchmarks cannot separate the effects of domain shift, lighting conditions, and anomaly size during evaluation. The proposed dataset allows controlled evaluation of these effects and supports evaluation of both closed-set and anomaly segmentation.