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

Benchmark Model Rank Results
domain-adaptation-on-gta5-to-cityscapesMRNet + Rectifying Label#22mIoU: 50.3
synthetic-to-real-translation-on-gtav-toMRNet+Rectifying Label#38mIoU: 50.3
synthetic-to-real-translation-on-synthia-to-1MRNet+Rectifying Label(ResNet-101)#22MIoU (16 classes): 47.9MIoU (13 classes): 54.9
unsupervised-domain-adaptation-on-cityscapes-2MRNet+Rectifying Label(ResNet-101)#2mIoU: 74.4
unsupervised-domain-adaptation-on-gtav-toUncertainty#16mIoU: 50.3
unsupervised-domain-adaptation-on-synthia-toUncertainty#17mIoU (13 classes): 54.9mIoU: 47.9