Scene Recognition on SUN397 base-to-new
83.7Base AccuracyTCP+CAKI
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| TCP+CAKIsetting=16-shot2026.05 | 83.7 | 78.2 | 80.9 | — | |
| GalLoP+CAKIsetting=16-shot2026.05 | 83.4 | 79.5 | 81.4 | — | |
| PromptSRC+CAKIsetting=16-shot2026.05 | 83.1 | 78.4 | 80.7 | — | |
| TCPsetting=16-shot2026.05 | 82.8 | 77.9 | 80.3 | — | |
| ProMetaRShots=162024.04 | 82.7 | 79.02 | 80.82 | — | |
| PromptSRCShots=162024.04 | 82.67 | 78.57 | 80.52 | — | |
| IVLPShots=162024.04 | 82.63 | 78.4 | 80.46 | — | |
| HPTBackbone=ViT-B/162023.12 | 82.57 | 79.26 | 80.88 | — | |
| GalLoPsetting=16-shot2026.05 | 82.5 | 78.4 | 80.4 | — | |
| PromptSRCsetting=16-shot2026.05 | 82.4 | 78.3 | 80.3 | — | |
| CoOpsetting=16-shot2026.05 | 81.1 | 68.4 | 74.2 | — | |
| MaPLeBackbone=ViT-B/162023.12 | 80.82 | 78.7 | 79.75 | — | |
| MaPLeShots=162024.04 | 80.82 | 78.7 | 79.75 | — | |
| CoOpshots=16, prompts=static learned, backbone=ViT-B/162022.03 | 80.6 | 65.89 | 72.51 | — | |
| CoOpShots=162024.04 | 80.6 | 65.89 | 72.51 | — | |
| RPOShots=162024.04 | 80.6 | 77.8 | 79.18 | — | |
| MaPLe+CAKIsetting=16-shot2026.05 | 80.6 | 78.8 | 79.7 | — | |
| MaPLesetting=16-shot2026.05 | 80.5 | 78.6 | 79.5 | — | |
| PromptSRC + CAKIShots=42026.05 | 80.5 | 78.4 | 79.4 | — | |
| UNIGRAMShots=162024.04 | 80.43 | 77.91 | 79.15 | — | |
| PromptSRCShots=42026.05 | 80.4 | 77.7 | 79 | — | |
| CoCoOpshots=16, prompts=dynamic conditional, backbone=ViT-B/162022.03 | 79.74 | 76.86 | 78.27 | — | |
| CoCoOpBackbone=ViT-B/162023.12 | 79.74 | 76.86 | 78.27 | — | |
| CoCoOpShots=162024.04 | 79.74 | 76.86 | 78.27 | — | |
| CoCoOpsetting=16-shot2026.05 | 79.6 | 77.1 | 78.3 | — | |
| MaPLe + CAKIShots=42026.05 | 79.2 | 78.3 | 78.8 | — | |
| MaPLeShots=42026.05 | 78.5 | 78.6 | 78.6 | — | |
| CoCoOp + CAKIShots=42026.05 | 76.7 | 78.4 | 77.5 | — | |
| GalLoP + CAKIShots=42026.05 | 76.4 | 57.5 | 65.6 | — | |
| CoCoOpShots=42026.05 | 75.7 | 78.5 | 77.1 | — | |
| GalLoPShots=42026.05 | 75.6 | 56.5 | 64.7 | — | |
| TCP + CAKIShots=42026.05 | 73.7 | 57.2 | 64.4 | — | |
| CoOp + CAKIShots=42026.05 | 73.2 | 72.6 | 72.9 | — | |
| TCPShots=42026.05 | 73.1 | 55.5 | 63.1 | — | |
| CoOpShots=42026.05 | 70.5 | 70.4 | 70.5 | — | |
| CLIPShots=42026.05 | 69.6 | 75.5 | 72.4 | — | |
| CLIPmode=zero-shot, prompts=manual, backbone=ViT-B/162022.03 | 69.36 | 75.35 | 72.23 | — | |
| CLIPBackbone=ViT-B/162023.12 | 69.36 | 75.35 | 72.23 | — | |
| CLIPShots=Zero-shot2024.04 | 69.36 | 75.35 | 72.23 | — | |
| CLIPProtocol=Base-to-novel generalization2025.12 | — | — | — | 72.23 | |
| CoCoOpProtocol=Base-to-novel generalization2025.12 | — | — | — | 78.27 | |
| CoOpProtocol=Base-to-novel generalization2025.12 | — | — | — | 74.07 | |
| FARLProtocol=Base-to-novel generalization2025.12 | — | — | — | 80.91 | |
| KgCoOpProtocol=Base-to-novel generalization2025.12 | — | — | — | 78.36 | |
| MaPLeProtocol=Base-to-novel generalization2025.12 | — | — | — | 79.75 | |
| MetaPromptProtocol=Base-to-novel generalization2025.12 | — | — | — | 80.62 | |
| MMAProtocol=Base-to-novel generalization2025.12 | — | — | — | 80.38 | |
| MMRLProtocol=Base-to-novel generalization2025.12 | — | — | — | 81.17 | |
| PLOTProtocol=Base-to-novel generalization2025.12 | — | — | — | 77.68 | |
| ProVPProtocol=Base-to-novel generalization2025.12 | — | — | — | 78.32 | |
| TCPProtocol=Base-to-novel generalization2025.12 | — | — | — | 80.35 |