The Fishyscapes Benchmark: Measuring Blind Spots in Semantic Segmentation

Benchmark Model Rank Results
anomaly-detection-on-fishyscapes-l-fDirichlet DeepLab#10AP: 34.28FPR95: 47.43
anomaly-detection-on-fishyscapes-l-fVoid Classifier#12AP: 10.29FPR95: 22.11
anomaly-detection-on-fishyscapes-l-fBayesian DeepLab#13AP: 9.8FPR95: 38.5
anomaly-detection-on-fishyscapes-l-fLearned Embedding Density#14AP: 4.7FPR95: 24.4
anomaly-detection-on-fishyscapes-l-fSoftmax Entropy#15AP: 2.9FPR95: 44.8