Image Classification on Occluded-OOD-CV 20-40% occlusion 1.0
62Top-1 Accuracy (Combined)UGT
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| UGTBackbone=ResNet-502024.03 | 62 | 62.4 | 60 | 72.6 | 63.2 | 64.4 | |
| DINEBackbone=ResNet-50, Pre-training=Pretrained ImageNet, Evaluation Protocol=Pretrained UDA model2024.03 | 60 | 51.5 | 42.3 | 57.2 | 61.8 | 52 | |
| MCCBackbone=ResNet-50, Pre-training=Pretrained ImageNet, Evaluation Protocol=Pretrained UDA model2024.03 | 58.2 | 57.7 | 50.3 | 67.1 | 61.1 | 60.1 | |
| MICBackbone=ResNet-50, Pre-training=Pretrained ImageNet, Evaluation Protocol=Pretrained UDA model2024.03 | 54 | 60.2 | 61.2 | 70.6 | 61.3 | 60.8 | |
| CDANBackbone=ResNet-50, Pre-training=Pretrained ImageNet, Evaluation Protocol=Pretrained UDA model2024.03 | 53.1 | 54.1 | 47.6 | 53.2 | 62 | 56.1 |