Align and Distill: Unifying and Improving Domain Adaptive Object Detection

Open paper
Benchmark Model Rank Results
unsupervised-domain-adaptation-on-cfc-daodALDI++ (ResNet50-FPN)#1AP@0.5: 76.1
unsupervised-domain-adaptation-on-cfc-daodMIC (ResNet50-FPN)#2AP@0.5: 74.1
unsupervised-domain-adaptation-on-cfc-daodAT (ResNet50-FPN)#3AP@0.5: 69.1
unsupervised-domain-adaptation-on-cfc-daodPT (ResNet50-FPN)#4AP@0.5: 69.0
unsupervised-domain-adaptation-on-cfc-daodUMT (ResNet50-FPN)#5AP@0.5: 61.2
unsupervised-domain-adaptation-on-cfc-daodSADA (ResNet50-FPN)#6AP@0.5: 58.9
unsupervised-domain-adaptation-on-cityscapes-to-foggy-cityscapesALDI-DETR (ResNet-50, 800px)#6mAP@0.5: 44.8
unsupervised-domain-adaptation-on-sim10k-to-cityscapesALDI++#1mAP@0.5: 77.8
unsupervised-domain-adaptation-on-sim10k-to-cityscapesALDI-YOLO#2mAP@0.5: 75.0
unsupervised-domain-adaptation-on-sim10k-to-cityscapesMIC(ALDI frame)#3mAP@0.5: 73.1
unsupervised-domain-adaptation-on-sim10k-to-cityscapesAT(ALDI frame)#4mAP@0.5: 72.0
unsupervised-domain-adaptation-on-sim10k-to-cityscapesSADA(ALDI frame)#5mAP@0.5: 71.8
unsupervised-domain-adaptation-on-sim10k-to-cityscapesPT(ALDI frame)#6mAP@0.5: 70.6

Rank counts only results with a code link.