Image Classification on Caltech101 Base and New Classes
98.97Base AccuracyMMRL
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MMRL2026.05 | 98.97 | 94.5 | 96.68 | |
| MMRL + NeRPRefining Method=NeRP2026.05 | 98.97 | 95.1 | 97 | |
| ManiPT2026.02 | 98.95 | 95.94 | 97.42 | |
| TAP2026.02 | 98.9 | 95.5 | 97.17 | |
| TAC2026.02 | 98.57 | 95.27 | 96.89 | |
| SkipT2026.05 | 98.5 | 95.33 | 96.89 | |
| LLaMP2026.02 | 98.45 | 95.85 | 97.13 | |
| MMAsetting=Base-to-Novel Generalization2025.12 | 98.4 | 94 | 96.15 | |
| RMAdaptersetting=Base-to-Novel Generalization2025.12 | 98.4 | 94.27 | 96.26 | |
| MMAsetting=Base-to-Novel Generalization2025.12 | 98.4 | 94 | 96.15 | |
| RMAdaptersetting=Base-to-Novel Generalization2025.12 | 98.4 | 94.27 | 96.26 | |
| MMA2026.05 | 98.4 | 94 | 96.15 | |
| MMA + NeRPRefining Method=NeRP2026.05 | 98.4 | 94.27 | 96.29 | |
| CoPrompt2023.06 | 98.27 | 94.9 | 96.55 | |
| PromptSRCsetting=Base-to-Novel Generalization2025.12 | 98.27 | 94 | 96.08 | |
| CoPromptsetting=Base-to-Novel Generalization2025.12 | 98.27 | 94.9 | 96.55 | |
| PromptSRCsetting=Base-to-Novel Generalization2025.12 | 98.27 | 94 | 96.08 | |
| CoPromptsetting=Base-to-Novel Generalization2025.12 | 98.27 | 94.9 | 96.55 | |
| CoPrompt2026.02 | 98.27 | 94.9 | 96.55 | |
| CoPrompt2026.05 | 98.27 | 94.9 | 96.55 | |
| TCPsetting=Base-to-Novel Generalization2025.12 | 98.23 | 94.67 | 96.42 | |
| TCPsetting=Base-to-Novel Generalization2025.12 | 98.23 | 94.67 | 96.42 | |
| CoOp + SEPTBackbone=CLIP, Evaluation Protocol=base-to-new generalization2026.01 | 98.23 | 93.4 | 95.76 | |
| MaPLeBackbone=frozen CLIP-ViT-B/16, Few-shot setting=16-shot, Number of random seeds=32026.05 | 98.23 | 95.1 | 96.64 | |
| MaPLeBackbone=CLIP-ViT-B/16, Training Shots=162026.05 | 98.23 | 95.1 | 96.64 | |
| MetaPromptMode=Meta-learning prompt2022.12 | 98.13 | 94.58 | 96.32 | |
| MetaPrompt2022.12 | 98.13 | 94.58 | 96.32 | |
| PromptSRC2023.06 | 98.1 | 94.03 | 96.02 | |
| MaPLesetting=Base-to-Novel Generalization2025.12 | 98.1 | 93.99 | 96.02 | |
| MaPLesetting=Base-to-Novel Generalization2025.12 | 98.1 | 93.99 | 96.02 | |
| PromptSRC2026.02 | 98.1 | 94.03 | 96.02 | |
| PromptSRC2026.05 | 98.1 | 94.03 | 96.02 | |
| PromptSRC + NeRPRefining Method=NeRP2026.05 | 98.1 | 94.47 | 96.25 | |
| CoOpBackbone=CLIP, Evaluation Protocol=base-to-new generalization2026.01 | 98.03 | 90.1 | 93.9 | |
| ProGrad2023.06 | 98.02 | 93.89 | 95.91 | |
| MaPLe2026.03 | 98.02 | 94.43 | 96.19 | |
| CoOpMode=Prompt learning2022.12 | 98 | 89.81 | 93.73 | |
| CoOp2022.12 | 98 | 89.81 | 93.73 | |
| CoOp2023.06 | 98 | 89.81 | 93.73 | |
| CoOpsetting=Base-to-Novel Generalization2025.12 | 98 | 89.81 | 93.73 | |
| CoOpsetting=Base-to-Novel Generalization2025.12 | 98 | 89.81 | 93.73 | |
| CoOp2026.02 | 98 | 89.81 | 93.73 | |
| BiMaPLeBackbone=frozen CLIP-ViT-B/16, Few-shot setting=16-shot, Number of random seeds=32026.05 | 98 | 94.8 | 96.37 | |
| BiMaPLeBackbone=CLIP-ViT-B/16, Training Shots=162026.05 | 98 | 94.8 | 96.37 | |
| CoOp2026.05 | 98 | 89.81 | 93.73 | |
| CoCoOpMode=Conditional prompt learning2022.12 | 97.96 | 93.81 | 95.84 | |
| CoCoOp2022.12 | 97.96 | 93.81 | 95.84 | |
| Co-CoOp2023.06 | 97.96 | 93.81 | 95.84 | |
| Co-CoOpsetting=Base-to-Novel Generalization2025.12 | 97.96 | 93.81 | 95.84 | |
| Co-CoOpsetting=Base-to-Novel Generalization2025.12 | 97.96 | 93.81 | 95.84 | |
| CoCoOp2026.02 | 97.96 | 93.81 | 95.84 | |
| CoCoOp2026.05 | 97.96 | 93.81 | 95.84 | |
| CoCoOp + NeRPRefining Method=NeRP2026.05 | 97.96 | 94.07 | 95.98 | |
| KgCoOpsetting=Base-to-Novel Generalization2025.12 | 97.92 | 93.99 | 95.91 | |
| KgCoOpsetting=Base-to-Novel Generalization2025.12 | 97.92 | 93.99 | 95.91 | |
| MaPLe + CC-CDFSL2026.03 | 97.91 | 95.49 | 96.68 | |
| MaPLeMode=Multi-modal prompt learning2022.12 | 97.74 | 94.36 | 96.02 | |
| MaPLe2022.12 | 97.74 | 94.36 | 96.02 | |
| MaPLe2023.06 | 97.74 | 94.36 | 96.02 | |
| MaPLe2026.02 | 97.74 | 94.36 | 96.02 | |
| MaPLe2026.05 | 97.74 | 94.36 | 96.02 | |
| KgCoOp2023.06 | 97.72 | 94.39 | 96.03 | |
| AdaptiveBiMaPLeBackbone=frozen CLIP-ViT-B/16, Few-shot setting=16-shot, Number of random seeds=32026.05 | 97.7 | 93.57 | 95.59 | |
| AdaptiveBiMaPLeBackbone=CLIP-ViT-B/16, Training Shots=16, Gate Learning Rate Multiplier=1x2026.05 | 97.7 | 93.57 | 95.59 | |
| AdaptiveBiMaPLe*Backbone=frozen CLIP-ViT-B/16, Few-shot setting=16-shot, Number of random seeds=3, Gate parameters learning rate=50x larger2026.05 | 97.67 | 93.6 | 95.59 | |
| AdaptiveBiMaPLeBackbone=CLIP-ViT-B/16, Training Shots=16, Gate Learning Rate Multiplier=50x2026.05 | 97.67 | 93.6 | 95.59 | |
| CLIPMode=Zero-shot2022.12 | 96.84 | 94 | 95.4 | |
| CLIP2022.12 | 96.84 | 94 | 95.4 | |
| CLIP2023.06 | 96.84 | 94 | 95.4 | |
| CLIPsetting=Base-to-Novel Generalization2025.12 | 96.84 | 94 | 95.4 | |
| CLIPsetting=Base-to-Novel Generalization2025.12 | 96.84 | 94 | 95.4 | |
| CLIP2026.02 | 96.84 | 94 | 95.4 |