Image Classification on EuroSAT Base-to-New
95.83Base ScoreVPT
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| VPTDePT=false2023.09 | 95.83 | 65.03 | 77.48 | |
| MMRL2026.05 | 95.6 | 80.17 | 87.21 | |
| MMRL + NeRPRefining Method=NeRP2026.05 | 95.6 | 85.17 | 90.08 | |
| PromptSRC + T_En2023.11 | 95.52 | 76.41 | 84.9 | |
| MaPLeDePT=false2023.09 | 94.67 | 66.73 | 78.28 | |
| CoPrompt2023.06 | 94.6 | 78.57 | 85.84 | |
| CoPromptProtocol=Base-to-Novel Generalization2025.12 | 94.6 | 78.57 | 85.84 | |
| CoPromptProtocol=Base-to-novel generalization2026.02 | 94.6 | 78.57 | 85.84 | |
| CoPrompt2026.05 | 94.6 | 78.57 | 85.84 | |
| 2SFSVisual Backbone=ViT-L/14, Shots (k)=82025.03 | 94.5 | 71.68 | 81.53 | |
| MaPLe + DePTDePT=true2023.09 | 94.43 | 76.23 | 84.36 | |
| TACProtocol=Base-to-novel generalization2026.02 | 94.37 | 82.6 | 88.1 | |
| MaPLemode=multi-modal-prompt-tuning2022.12 | 94.07 | 73.23 | 82.35 | |
| MaPLe2022.12 | 94.07 | 73.23 | 82.35 | |
| MaPLe2023.06 | 94.07 | 73.23 | 82.35 | |
| MaPLe2024.12 | 94.07 | 73.23 | 82.35 | |
| MaPLeProtocol=Base-to-Novel Generalization2025.12 | 94.07 | 73.23 | 82.3 | |
| MaPLeProtocol=Base-to-novel generalization2026.02 | 94.07 | 73.23 | 82.35 | |
| MaPLe2026.05 | 94.07 | 73.23 | 82.35 | |
| ManiPTProtocol=Base-to-novel generalization2026.02 | 94.06 | 83.72 | 88.59 | |
| VPT + DePTDePT=true2023.09 | 93.77 | 76.3 | 84.14 | |
| CoPro.*2024.12 | 93.77 | 71.73 | 81.28 | |
| MaPLe + CC-CDFSL2026.03 | 93.57 | 71.98 | 81.37 | |
| MetaPromptmode=meta-prompt-learning2022.12 | 93.53 | 75.21 | 83.38 | |
| MetaPrompt2022.12 | 93.53 | 75.21 | 83.38 | |
| PromptSRC + DePTDePT=true2023.09 | 93.23 | 77.9 | 84.88 | |
| PromptSRC2023.06 | 92.9 | 73.9 | 82.32 | |
| ProSRC.2024.12 | 92.9 | 73.9 | 82.32 | |
| PromptSRCProtocol=Base-to-Novel Generalization2025.12 | 92.9 | 73.9 | 82.32 | |
| PromptSRCProtocol=Base-to-novel generalization2026.02 | 92.9 | 73.9 | 82.32 | |
| PromptSRC2026.05 | 92.9 | 73.9 | 82.32 | |
| PromptSRC + NeRPRefining Method=NeRP2026.05 | 92.9 | 78.53 | 85.11 | |
| PromptSRCDePT=false2023.09 | 92.8 | 72.07 | 81.13 | |
| MaPLe2026.03 | 92.66 | 68.39 | 78.7 | |
| SkipT.2024.12 | 92.47 | 83 | 87.48 | |
| SkipT2026.05 | 92.47 | 83 | 87.48 | |
| CoOpmode=prompt-tuning2022.12 | 92.19 | 54.74 | 68.69 | |
| CoOp2022.12 | 92.19 | 54.74 | 68.69 | |
| CoOp2023.06 | 92.19 | 54.74 | 68.69 | |
| CoOp2024.12 | 92.19 | 54.74 | 68.69 | |
| CoOpProtocol=Base-to-Novel Generalization2025.12 | 92.19 | 54.74 | 68.69 | |
| CoOpProtocol=Base-to-novel generalization2026.02 | 92.19 | 54.74 | 68.69 | |
| CoOp2026.05 | 92.19 | 54.74 | 68.69 | |
| RMAdapterProtocol=Base-to-Novel Generalization2025.12 | 92.17 | 78.33 | 84.69 | |
| LLaMPProtocol=Base-to-novel generalization2026.02 | 91.93 | 83.66 | 87.6 | |
| TCP2024.12 | 91.63 | 74.73 | 82.32 | |
| TCPProtocol=Base-to-Novel Generalization2025.12 | 91.63 | 74.73 | 82.32 | |
| TAPProtocol=Base-to-novel generalization2026.02 | 90.7 | 82.17 | 86.22 | |
| CoCoOp + DePTDePT=true2023.09 | 90.27 | 66.87 | 76.82 | |
| ProGrad2023.06 | 90.11 | 60.89 | 72.67 | |
| ProGrad2024.12 | 90.11 | 60.89 | 72.67 | |
| CoOpDePT=false2023.09 | 89.35 | 57.3 | 69.82 | |
| KgCoOp + DePTDePT=true2023.09 | 89.03 | 71.07 | 79.04 | |
| DePT2024.12 | 89.03 | 71.07 | 79.04 | |
| CoOp + DePTDePT=true2023.09 | 88.27 | 66.27 | 75.7 | |
| CoCoOpDePT=false2023.09 | 87.97 | 63.63 | 73.85 | |
| CoCoOpmode=conditional-prompt-tuning2022.12 | 87.49 | 60.04 | 71.21 | |
| CoCoOp2022.12 | 87.49 | 60.04 | 71.21 | |
| Co-CoOp2023.06 | 87.49 | 60.04 | 71.21 | |
| CoCoOp2024.12 | 87.49 | 60.04 | 71.21 | |
| Co-CoOpProtocol=Base-to-Novel Generalization2025.12 | 87.49 | 60.04 | 71.21 | |
| CoCoOpProtocol=Base-to-novel generalization2026.02 | 87.49 | 60.04 | 71.21 | |
| CoCoOp2026.05 | 87.49 | 60.04 | 71.21 | |
| CoCoOp + NeRPRefining Method=NeRP2026.05 | 87.49 | 67.3 | 76.08 | |
| KgCoOpDePT=false2023.09 | 85.77 | 63.4 | 72.91 | |
| AdaptiveBiMaPLeBackbone=CLIP-ViT-B/16, Shots=16-shot, Gate Learning Rate=1x2026.05 | 85.73 | 68.48 | 76.14 | |
| AdaptiveBiMaPLe*Backbone=CLIP-ViT-B/16, Shots=16-shot, Gate Learning Rate=50x2026.05 | 85.73 | 68.8 | 76.33 | |
| AdaptiveBiMaPLeBackbone=CLIP-ViT-B/16, Training Shots=16, Gate Learning Rate Multiplier=1x2026.05 | 85.73 | 68.48 | 76.14 | |
| AdaptiveBiMaPLeBackbone=CLIP-ViT-B/16, Training Shots=16, Gate Learning Rate Multiplier=50x2026.05 | 85.73 | 68.8 | 76.33 | |
| KgCoOp2023.06 | 85.64 | 64.34 | 73.48 | |
| KgCoOp2024.12 | 85.64 | 64.34 | 73.48 | |
| KgCoOpProtocol=Base-to-Novel Generalization2025.12 | 85.64 | 64.34 | 73.48 | |
| MMAProtocol=Base-to-Novel Generalization2025.12 | 85.46 | 82.34 | 83.87 | |
| MMA2026.05 | 85.46 | 82.34 | 83.87 | |
| MMA + NeRPRefining Method=NeRP2026.05 | 85.46 | 85.27 | 85.36 | |
| MaPLeBackbone=CLIP-ViT-B/16, Shots=16-shot2026.05 | 83.87 | 57.65 | 68.33 | |
| MaPLeBackbone=CLIP-ViT-B/16, Training Shots=162026.05 | 83.87 | 57.65 | 68.33 | |
| BiMaPLeBackbone=CLIP-ViT-B/16, Shots=16-shot2026.05 | 82.68 | 63.08 | 71.56 | |
| BiMaPLeBackbone=CLIP-ViT-B/16, Training Shots=162026.05 | 82.68 | 63.08 | 71.56 | |
| MMAVisual Backbone=ViT-L/14, Shots (k)=82025.03 | 72.73 | 72 | 72.36 | |
| CLIPVisual Backbone=ViT-L/14, Shots (k)=82025.03 | 70.93 | 82.9 | 76.45 | |
| CLIPmode=zero-shot2022.12 | 56.48 | 64.05 | 60.03 | |
| CLIPmode=zero-shot2022.12 | 56.48 | 64.05 | 60.03 | |
| CLIPVision Encoder=ViT-B/162023.11 | 56.48 | 64.05 | 60.03 | |
| CLIP2023.06 | 56.48 | 64.05 | 60.03 | |
| CLIPProtocol=Base-to-Novel Generalization2025.12 | 56.48 | 64.05 | 60.03 | |
| CLIPProtocol=Base-to-novel generalization2026.02 | 56.48 | 64.05 | 60.03 |