Image Classification on PatternNet
99.67AccuracyArcGate
Evaluation Results
| Method | Links | |
|---|---|---|
| ArcGateBackbone=ResNet-502026.05 | 99.67 | |
| ReLUBackbone=ResNet-502026.05 | 99.57 | |
| Leaky-ReLUBackbone=ResNet-502026.05 | 99.57 | |
| ArcGateBackbone=ViT-B/162026.05 | 99.48 | |
| GELUBackbone=ViT-B/162026.05 | 99.39 | |
| ReLUBackbone=ViT-B/162026.05 | 99.19 | |
| Leaky-ReLUBackbone=ViT-B/162026.05 | 99.18 | |
| GELUBackbone=ResNet-502026.05 | 99.18 | |
| SiLU (Swish)Backbone=ResNet-502026.05 | 99.01 | |
| TanhBackbone=ResNet-502026.05 | 98.27 | |
| SiLU (Swish)Backbone=ViT-B/162026.05 | 96.99 | |
| SOTASetting=Transductive, Transductive Inference (TransCLIP)=false2025.06 | 96.5 | |
| SigmoidBackbone=ViT-B/162026.05 | 95.61 | |
| T-GeoRSCLIPSetting=Transductive, Transductive Inference (TransCLIP)=true2025.06 | 94.5 | |
| TanhBackbone=ViT-B/162026.05 | 92.89 | |
| AVIONBackbone=GeoRSCLIP (ViT-B/32), Evaluation Protocol=16-shot2026.03 | 92.09 | |
| T-SkyCLIP50Setting=Transductive, Transductive Inference (TransCLIP)=true2025.06 | 87.6 | |
| APPLeNetBackbone=GeoRSCLIP (ViT-B/32), Evaluation Protocol=16-shot2026.03 | 86.63 | |
| MetaCLIP-2.5BArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 82.8 | |
| GeoRSCLIPArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 80.9 | |
| SkyCLIPArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 80.6 | |
| SigLIP2Architecture=ViT-L/16, Setting=Zero-shot, Template=DST2026.06 | 78.9 | |
| CLIP LAION-RSArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 77.8 | |
| T-CLIPSetting=Transductive, Transductive Inference (TransCLIP)=true2025.06 | 76.6 | |
| GeoRSCLIPArchitecture=ViT-B/32, Setting=Zero-shot, Template=DST2026.06 | 76.6 | |
| CLIPArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 75.9 | |
| GeoRSCLIPSetting=Transductive, Transductive Inference (TransCLIP)=false2025.06 | 75.8 | |
| MetaCLIP-400MArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 74.4 | |
| SkyCLIP50Setting=Transductive, Transductive Inference (TransCLIP)=false2025.06 | 73.8 | |
| CLIPBackbone=ViT-L-14, zero-shot=true2023.06 | 73.78 | |
| CLIPBackbone=ViT-L-14, zero-shot=true2023.06 | 73.78 | |
| MetaCLIP-2.5BArchitecture=ViT-B/16, Setting=Zero-shot, Template=DST2026.06 | 71.9 | |
| SigLIP2Architecture=ViT-B/16, Setting=Zero-shot, Template=DST2026.06 | 71 | |
| T-RemoteCLIPSetting=Transductive, Transductive Inference (TransCLIP)=true2025.06 | 70.7 | |
| CLIPBackbone=ViT-L, zero-shot=true2025.12 | 70.15 | |
| MetaCLIP-400MArchitecture=ViT-B/16, Setting=Zero-shot, Template=DST2026.06 | 69.1 | |
| RemoteCLIPBackbone=ViT-L-14, zero-shot=true2023.06 | 68.75 | |
| RemoteCLIPBackbone=ViT-L-14, zero-shot=true2023.06 | 68.75 | |
| SigLIPArchitecture=ViT-L/16, Setting=Zero-shot, Template=DST2026.06 | 65.5 | |
| CLIPArchitecture=ViT-B/16, Setting=Zero-shot, Template=DST2026.06 | 64.5 | |
| SigLIPArchitecture=ViT-B/16, Setting=Zero-shot, Template=DST2026.06 | 64.2 | |
| RemoteCLIPArchitecture=ViT-L/14, Setting=Zero-shot, Template=DST2026.06 | 64.1 | |
| GeoMELTBackbone=BEiT-3, zero-shot=true2025.12 | 64.01 | |
| RemoteCLIPBackbone=ViT-L, zero-shot=true2025.12 | 63.12 | |
| GeoRSCLIPBackbone=GeoRSCLIP (ViT-B/32), Evaluation Protocol=Zero-shot2026.03 | 60.17 | |
| CLIPSetting=Transductive, Transductive Inference (TransCLIP)=false2025.06 | 59.6 | |
| RemoteCLIPArchitecture=ViT-B/32, Setting=Zero-shot, Template=DST2026.06 | 59.3 | |
| CLIPBackbone=ViT-B-32, zero-shot=true2023.06 | 58.95 | |
| CLIPBackbone=ViT-B-32, zero-shot=true2023.06 | 58.95 | |
| RemoteCLIPBackbone=ViT-B, zero-shot=true2025.12 | 58.01 | |
| RemoteCLIPBackbone=ViT-B-32, zero-shot=true2023.06 | 57.71 | |
| RemoteCLIPBackbone=ViT-B-32, zero-shot=true2023.06 | 57.71 | |
| RemoteCLIPSetting=Transductive, Transductive Inference (TransCLIP)=false2025.06 | 55.9 | |
| CLIPBackbone=ViT-B, zero-shot=true2025.12 | 49.65 | |
| CLIPBackbone=ResNet-50, zero-shot=true2023.06 | 46.96 | |
| CLIPBackbone=ResNet-50, zero-shot=true2023.06 | 46.96 | |
| RemoteCLIPArchitecture=RN-50, Setting=Zero-shot, Template=DST2026.06 | 45.9 | |
| RemoteCLIPBackbone=ResNet-50, zero-shot=true2023.06 | 45.51 | |
| RemoteCLIPBackbone=ResNet-50, zero-shot=true2023.06 | 45.51 | |
| Domain-Aware Reinforcement Fine-TuningBase Model=QwenVL, Fine-tuning Strategy=Reinforcement Fine-Tuning2026.01 | 19.1 | |
| SigmoidBackbone=ResNet-502026.05 | 16.63 | |
| QwenVL + Domain PromptBase Model=QwenVL, Injection Strategy=Domain Prompt2026.01 | 8.71 | |
| QwenVLBase Model=QwenVL2026.01 | 8.62 | |
| QwenVL + Caption(BLIP)Base Model=QwenVL, Injection Strategy=Caption (BLIP)2026.01 | 8.52 | |
| QwenVL + Caption(MLLM)Base Model=QwenVL, Injection Strategy=Caption (MLLM)2026.01 | 5.23 |