Classification on m-so2sat (test)
69.8Mean AccuracyOlmoEarth v1.1
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| OlmoEarth v1.1ViT Size=ViT Base, Modalities=S2, Time series=false2026.05 | 69.8 | — | — | — | |
| CopernicusFMViT Size=ViT Base, Modalities=S2, Time series=false2026.05 | 66.8 | — | — | — | |
| ClayViT Size=ViT Large, Modalities=S2, Time series=false2026.05 | 61.3 | — | — | — | |
| DOFA-ViT-300MBackbone=ViT, Parameters=300M, Pre-training=DOFA2024.12 | 61.26 | 1.97 | 63.89 | 57.81 | |
| Prithvi-EO-2.0-600M-TLVersion=2.0, Parameters=600M, Transfer Learning=true2024.12 | 59.54 | 2.22 | 64 | 55.58 | |
| Prithvi-EO-2.0-300MVersion=2.0, Parameters=300M2024.12 | 58.44 | 1.33 | 60.45 | 56.49 | |
| Prithvi-EO-2.0-300M-TLVersion=2.0, Parameters=300M, Transfer Learning=true2024.12 | 57.84 | 1.31 | 59.13 | 54.46 | |
| Prithvi-EO-2.0-600MVersion=2.0, Parameters=600M2024.12 | 57.2 | 2.17 | 61.46 | 54.16 | |
| Prithvi-EO-2.0-100M-TLVersion=2.0, Parameters=100M, Transfer Learning=true2024.12 | 56.8 | 1.28 | 58.22 | 54.67 | |
| Satlas-Swin-100MBackbone=Swin, Parameters=100M, Pre-training=Satlas2024.12 | 56.74 | 2.7 | 58.82 | 50.51 | |
| AnysatViT Size=ViT Base, Modalities=S2, Time series=false2026.05 | 56.7 | — | — | — | |
| DeCUR-Resnet50Backbone=ResNet-50, Pre-training=DeCUR2024.12 | 56.68 | 1.78 | 58.32 | 53.65 | |
| Prithvi-EO-2.0-100MVersion=2.0, Parameters=100M2024.12 | 55.59 | 1.31 | 57.4 | 53.85 | |
| MOCO-Resnet50Backbone=ResNet-50, Pre-training=MOCO2024.12 | 55.58 | 1.87 | 59.74 | 53.25 | |
| DINO-Resnet50Backbone=ResNet-50, Pre-training=DINO2024.12 | 54.39 | 1.94 | 56.69 | 51.22 | |
| Prithvi-EO-1.0-100MVersion=1.0, Parameters=100M2024.12 | 51.5 | 1.55 | 54.16 | 49.59 | |
| ScaleMAE-ViT-300MBackbone=ViT, Parameters=300M, Pre-training=ScaleMAE2024.12 | 48.33 | 0.79 | 49.29 | 47.16 |