Fine-grained classification on Stanford Cars (base classes)
0.91MCECoOp
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CoOpPrompt-tuning strategy=CoOp2026.02 | 0.91 | 3.65 | |
| TCPT (Ours)Prompt-tuning strategy=KGCoOp2026.02 | 1.12 | 8.1 | |
| TCPT (Ours)Prompt-tuning strategy=MaPLe2026.02 | 1.2 | 3.98 | |
| MaPLePrompt-tuning strategy=MaPLe2026.02 | 1.68 | 6.91 | |
| TCPT (Ours)Prompt-tuning strategy=CoOp2026.02 | 1.87 | 7.55 | |
| PenaltyPrompt-tuning strategy=KGCoOp2026.02 | 2.1 | 9.96 | |
| PenaltyPrompt-tuning strategy=CoOp2026.02 | 2.13 | 10.01 | |
| ZS-NormPrompt-tuning strategy=CoOp2026.02 | 2.2 | 11.26 | |
| KGCoOpPrompt-tuning strategy=KGCoOp2026.02 | 2.23 | 10.14 | |
| ZS-NormPrompt-tuning strategy=KGCoOp2026.02 | 2.28 | 10.58 | |
| PenaltyPrompt-tuning strategy=MaPLe2026.02 | 2.3 | 11.17 | |
| ZS-NormPrompt-tuning strategy=MaPLe2026.02 | 3.79 | 20.73 |