Image Classification on AID
96.6AccuracySOTA
Evaluation Results
| Method | Links | |
|---|---|---|
| SOTASetting=Transductive, Transductive Inference (TransCLIP)=false2025.06 | 96.6 | |
| T-RemoteCLIPSetting=Transductive, Transductive Inference (TransCLIP)=true2025.06 | 95.6 | |
| DINOv2-LVD142MBackbone=ViT-L/14, Training Scale=large scale datasets, Evaluation protocol=Linear probe2026.06 | 94.27 | |
| Earth-Agent2025.09 | 93.42 | |
| S-CLIPModel=ViT-B/162023.05 | 93.1 | |
| VISReg-Inet22KBackbone=ViT-L/14, Training Scale=large scale datasets, Evaluation protocol=Linear probe2026.06 | 92.81 | |
| S-CLIPModel=ViT-B/322023.05 | 92.5 | |
| CLIPModel=ViT-B/32, Training Mode=fine-tune2023.05 | 92 | |
| iBOTBackbone=ViT-B/16, Heuristics=Included, Evaluation protocol=Linear probe2026.06 | 91.73 | |
| VHM2025.09 | 91.7 | |
| RemoteCLIPSetting=Transductive, Transductive Inference (TransCLIP)=false2025.06 | 91.7 | |
| DINOBackbone=ViT-B/16, Heuristics=Included, Evaluation protocol=Linear probe2026.06 | 91.52 | |
| RemoteCLIPArchitecture=ViT-B/32, Setting=Zero-shot, Template=DST2026.06 | 91.5 | |
| RemoteCLIPBackbone=ViT-B-32, zero-shot=true2023.06 | 91.3 | |
| RemoteCLIPBackbone=ViT-B-32, zero-shot=true2023.06 | 91.3 | |
| LHRS-Bot2025.09 | 91.26 | |
| iBOTBackbone=ViT-L/16, Heuristics=Included, Evaluation protocol=Linear probe2026.06 | 91.24 | |
| VISRegBackbone=ViT-B/16, Heuristics=Excluded, Evaluation protocol=Linear probe2026.06 | 90.91 | |
| MoCoV3Backbone=ViT-B/16, Heuristics=Included, Evaluation protocol=Linear probe2026.06 | 90.2 | |
| VISRegBackbone=ViT-L/14, Heuristics=Excluded, Evaluation protocol=Linear probe2026.06 | 90.12 | |
| RemoteCLIPArchitecture=RN-50, Setting=Zero-shot, Template=DST2026.06 | 89.4 | |
| CLIPModel=ViT-B/16, Training Mode=fine-tune2023.05 | 89.2 | |
| EarthDialBackbone=Phi-3, zero-shot=true2025.12 | 88.76 | |
| I-JEPABackbone=ViT-H/14, Heuristics=Included, Evaluation protocol=Linear probe2026.06 | 88.68 | |
| GeoVLM-R1Backbone=Qwen2.5VL-3B, zero-shot=true2025.12 | 88.46 | |
| RemoteCLIPBackbone=ViT-L-14, zero-shot=true2023.06 | 87.9 | |
| RemoteCLIPBackbone=ViT-L-14, zero-shot=true2023.06 | 87.9 | |
| S-CLIPModel=ResNet-502023.05 | 87.7 | |
| GeoMELTBackbone=BEiT-3, zero-shot=true2025.12 | 87.4 | |
| RemoteCLIPBackbone=ViT-B, zero-shot=true2025.12 | 86.6 | |
| RemoteCLIPBackbone=ResNet-50, zero-shot=true2023.06 | 86.55 | |
| RemoteCLIPBackbone=ResNet-50, zero-shot=true2023.06 | 86.55 | |
| MAEBackbone=ViT-L/16, Heuristics=Excluded, Evaluation protocol=Linear probe2026.06 | 86.42 | |
| data2vecBackbone=ViT-L/16, Heuristics=Included, Evaluation protocol=Linear probe2026.06 | 85.98 | |
| SkyNativeParams.=8B2026.05 | 84.83 | |
| RemoteCLIPArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 84.2 | |
| CLIPModel=ResNet-50, Training Mode=fine-tune2023.05 | 83.6 | |
| T-CLIPSetting=Transductive, Transductive Inference (TransCLIP)=true2025.06 | 80.7 | |
| VHMParams.=7B2026.05 | 79 | |
| MetaCLIP-2.5BArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 78.9 | |
| T-SkyCLIP50Setting=Transductive, Transductive Inference (TransCLIP)=true2025.06 | 78.7 | |
| T-GeoRSCLIPSetting=Transductive, Transductive Inference (TransCLIP)=true2025.06 | 78.2 | |
| AVIONBackbone=GeoRSCLIP (ViT-B/32), Evaluation Protocol=Zero-shot2026.03 | 77.99 | |
| RemoteCLIPBackbone=ViT-L, zero-shot=true2025.12 | 76.95 | |
| GeoRSCLIPArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 76.9 | |
| APPLeNetBackbone=GeoRSCLIP (ViT-B/32), Evaluation Protocol=Zero-shot2026.03 | 76.4 | |
| Gemini-2.02026.05 | 76 | |
| SkySenseGPTParams.=7B2026.05 | 75.5 | |
| SigLIP2Architecture=ViT-L/16, Setting=Zero-shot, Template=DST2026.06 | 74.7 | |
| MetaCLIP-400MArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 73.9 | |
| GeoRSCLIPBackbone=GeoRSCLIP (ViT-B/32), Evaluation Protocol=Zero-shot2026.03 | 73.72 | |
| SigLIP2Architecture=ViT-B/16, Setting=Zero-shot, Template=DST2026.06 | 72.7 | |
| MetaCLIP-2.5BArchitecture=ViT-B/16, Setting=Zero-shot, Template=DST2026.06 | 72.5 | |
| CLIP LAION-RSArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 72.5 | |
| SigLIPArchitecture=ViT-L/16, Setting=Zero-shot, Template=DST2026.06 | 72.2 | |
| SkyCLIPArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 72.2 | |
| GeoRSCLIPArchitecture=ViT-B/32, Setting=Zero-shot, Template=DST2026.06 | 72.1 | |
| GeoChatBackbone=LLaVA-1.5-7B, zero-shot=true2025.12 | 72.03 | |
| Geochat2025.09 | 72.03 | |
| MetaCLIP-400MArchitecture=ViT-B/16, Setting=Zero-shot, Template=DST2026.06 | 71.8 | |
| SigLIPArchitecture=ViT-B/16, Setting=Zero-shot, Template=DST2026.06 | 71 | |
| GeoRSCLIPSetting=Transductive, Transductive Inference (TransCLIP)=false2025.06 | 70.3 | |
| SkyCLIP50Setting=Transductive, Transductive Inference (TransCLIP)=false2025.06 | 70.3 | |
| CLIPBackbone=ViT-L-14, zero-shot=true2023.06 | 69.88 | |
| CLIPBackbone=ViT-L-14, zero-shot=true2023.06 | 69.88 | |
| CLIPArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 69.4 | |
| Qwen2.5-VLParams.=7B2026.05 | 68.17 | |
| EarthDialParams.=4B2026.05 | 67.33 | |
| CLIPArchitecture=ViT-B/16, Setting=Zero-shot, Template=DST2026.06 | 67.3 | |
| CLIPSetting=Transductive, Transductive Inference (TransCLIP)=false2025.06 | 66.4 | |
| CLIPBackbone=ViT-B-32, zero-shot=true2023.06 | 65.65 | |
| CLIPBackbone=ViT-B-32, zero-shot=true2023.06 | 65.65 | |
| EVEv2Params.=7B2026.05 | 64.17 | |
| CLIPBackbone=ViT-L, zero-shot=true2025.12 | 62.7 | |
| Claude-42026.05 | 60.33 | |
| Sphinx2025.09 | 58.2 | |
| CLIPBackbone=ResNet-50, zero-shot=true2023.06 | 57.35 | |
| CLIPBackbone=ResNet-50, zero-shot=true2023.06 | 57.35 | |
| CLIPBackbone=ViT-B, zero-shot=true2025.12 | 51.4 | |
| LLaVA-1.52025.09 | 51 | |
| VoCoTParams.=7B2026.05 | 46.67 | |
| MiniGPT-v22025.09 | 32.96 | |
| MiniGPT-v2Params.=7B2026.05 | 22.5 |