Rectifying Pseudo Label Learning via Uncertainty Estimation for Domain Adaptive Semantic Segmentation

Benchmark Model Rank Results
unsupervised-domain-adaptation-on-cityscapes-to-oxfordcarMRNet+Rectifying Label(ResNet-101)#2mIoU: 74.4
unsupervised-domain-adaptation-on-synthia-to-cityscapesUncertainty#17mIoU (13 classes): 54.9mIoU: 47.9
unsupervised-domain-adaptation-on-gtav-to-cityscapes-labelsUncertainty#21mIoU: 50.3
synthetic-to-real-translation-on-synthia-to-cityscapesMRNet+Rectifying Label(ResNet-101)#26MIoU (16 classes): 47.9MIoU (13 classes): 54.9
domain-adaptation-on-gta5-to-cityscapesMRNet + Rectifying Label#27mIoU: 50.3
synthetic-to-real-translation-on-gtav-to-cityscapes-labelsMRNet+Rectifying Label#44mIoU: 50.3