Land Cover Classification on UC-Merced
99.48Top-1 AccuracyCMID
Evaluation results
| Method | Links | |
|---|---|---|
| CMIDGeneral pre-training=14.0M (ImageNet-21k), Remote sensing pre-training=1.00M (MillionAID), Parameters=31M (Swin-B)2026.05 | 99.48 | |
| RingMo-LiteRemote sensing pre-training=0.15M, Parameters=28M (Swin-T)2026.05 | 99.1 | |
| SwinGeneral pre-training=14.0M (ImageNet-21k), Parameters=88M (Swin-B)2026.05 | 99 | |
| SatlasNetGeneral pre-training=14.0M (ImageNet-21k), Remote sensing pre-training=0.90M (SatlasPretrain), Parameters=88M (Swin-B)2026.05 | 99 | |
| GFMRemote sensing pre-training=0.60M (GeoPile), Parameters=88M (Swin-B)2026.05 | 99 | |
| ResNetGeneral pre-training=1.3M (ImageNet-1k), Parameters=26M (ResNet-50)2026.05 | 98.8 | |
| Foundation ModelRemote sensing pre-training=0.28M, Parameters=32M (Swin-T)2026.05 | 98.1 | |
| SatMAE++Backbone=ViT-Large, Protocol=Finetuning2024.03 | 97.62 | |
| SatMAE++General pre-training=14.0M (ImageNet-21k), Remote sensing pre-training=0.70M (fMoW-Sentinel), Parameters=307M (ViT-L)2026.05 | 97.6 | |
| SeCoGeneral pre-training=1.3M (ImageNet-1k), Remote sensing pre-training=1.00M (SeCo), Parameters=26M (ResNet-50)2026.05 | 97.1 | |
| SatMAEGeneral pre-training=14.0M (ImageNet-21k), Remote sensing pre-training=0.70M (fMoW-Sentinel), Parameters=307M (ViT-L)2026.05 | 94.1 | |
| SatMAEBackbone=ViT-Large, Protocol=Finetuning2024.03 | 94.05 | |
| ViTGeneral pre-training=14.0M (ImageNet-21k), Parameters=307M (ViT-L)2026.05 | 93.1 |