Classification on 12 Remote Sensing Datasets Average (test)
93.93AccuracyRemoteCLIP
Evaluation results
| Method | Links | |
|---|---|---|
| RemoteCLIPBackbone=ViT-Base, Evaluation Protocol=Linear probing2023.06 | 93.93 | |
| RemoteCLIPBackbone=ViT-Base, Evaluation Protocol=k-NN2023.06 | 93.77 | |
| CLIPBackbone=ViT-Base, Evaluation Protocol=Linear probing2023.06 | 92.31 | |
| CLIPBackbone=ViT-Base, Evaluation Protocol=k-NN2023.06 | 92.15 | |
| RemoteCLIPBackbone=ResNet-50, Evaluation Protocol=Linear probing2023.06 | 92.06 | |
| RemoteCLIPBackbone=ResNet-50, Evaluation Protocol=k-NN2023.06 | 92.04 | |
| CLIP-CLBackbone=ResNet-50, Evaluation Protocol=k-NN2023.06 | 91.25 | |
| CLIP-CLBackbone=ResNet-50, Evaluation Protocol=Linear probing2023.06 | 91.18 | |
| Barlow TwinsBackbone=ResNet-50, Evaluation Protocol=Linear probing2023.06 | 90 | |
| VICRegBackbone=ResNet-50, Evaluation Protocol=Linear probing2023.06 | 89.85 | |
| Barlow TwinsBackbone=ResNet-50, Evaluation Protocol=k-NN2023.06 | 89.73 | |
| VICRegBackbone=ResNet-50, Evaluation Protocol=k-NN2023.06 | 89.56 | |
| CLIPBackbone=ResNet-50, Evaluation Protocol=Linear probing2023.06 | 89.47 | |
| CLIPBackbone=ResNet-50, Evaluation Protocol=k-NN2023.06 | 89.42 | |
| SwAVBackbone=ResNet-50, Evaluation Protocol=Linear probing2023.06 | 88.97 | |
| SwAVBackbone=ResNet-50, Evaluation Protocol=k-NN2023.06 | 88.83 | |
| ImageNetBackbone=ResNet-50, Evaluation Protocol=Linear probing2023.06 | 87.32 | |
| ImageNetBackbone=ResNet-50, Evaluation Protocol=k-NN2023.06 | 87.08 | |
| ImageNetBackbone=ViT-Base, Evaluation Protocol=Linear probing2023.06 | 86.34 | |
| ImageNetBackbone=ViT-Base, Evaluation Protocol=k-NN2023.06 | 86.05 | |
| ViTAEBackbone=ViTAE, Evaluation Protocol=Linear probing2023.06 | 84.02 | |
| ViTAEBackbone=ViTAE, Evaluation Protocol=k-NN2023.06 | 83.03 |