Adaptive Boosting for Domain Adaptation: Towards Robust Predictions in Scene Segmentation

Benchmark Model Rank Results
domain-adaptation-on-gta5-synscapes-toMRNet + Adaboost#2mIoU: 50.8
domain-adaptation-on-gta5-to-cityscapesMRNet + Adaboost#24mIoU: 49.0
domain-adaptation-on-gtav-synscapes-toMRNet+Adaboost#3mIoU: 50.8
semi-supervised-image-classification-on-cifarAdaboost#19Percentage error: 6.05±0.12
synthetic-to-real-translation-on-gtav-toUncertainty + Adaboost#37mIoU: 50.9
synthetic-to-real-translation-on-synthia-to-1Uncertainty + Adaboost (ResNet-101)#19MIoU (16 classes): 50.4MIoU (13 classes): 57.5
synthetic-to-real-translation-on-synthia-to-1MRNet + Adaboost (ResNet-101)#25MIoU (16 classes): 45.9MIoU (13 classes): 52.9
unsupervised-domain-adaptation-on-cityscapes-2Uncertainty + Adaboost#1mIoU: 75.2
unsupervised-domain-adaptation-on-cityscapes-2MRNet + Adaboost#4mIoU: 73.7
unsupervised-domain-adaptation-on-gtav-toUncertainty + Adaboost#15mIoU: 50.9
unsupervised-domain-adaptation-on-synthia-toUncertainty + Adaboost#16mIoU (13 classes): 57.5mIoU: 50.4
unsupervised-domain-adaptation-on-synthia-toMRNet + Adaboost#18mIoU (13 classes): 52.9mIoU: 45.9