Fine-grained Image Classification on Caltech101
94.44Top-1 AccuracyRLCF
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| RLCFmode=Test-Time Adaptation2026.02 | 94.44 | — | — | |
| MTAmode=Test-Time Adaptation2026.02 | 94.4 | — | — | |
| TPTmode=Test-Time Adaptation2026.02 | 94.24 | — | — | |
| TPSmode=Test-Time Adaptation2026.02 | 94.16 | — | — | |
| CLIPmode=Zero-shot2026.02 | 94 | — | — | |
| ZEROmode=Test-Time Adaptation2026.02 | 94 | — | — | |
| R-TPTmode=Test-Time Adaptation2026.02 | 93.91 | — | — | |
| TTLmode=Test-Time Adaptation2026.02 | 93.75 | — | — | |
| C-TPTmode=Test-Time Adaptation2026.02 | 93.71 | — | — | |
| Fair Context Learning (FCL)mode=Test-Time Adaptation2026.02 | 91.76 | — | — | |
| C-TPTBackbone=CLIP-ViT-B/16, Epsilon (ε)=4.02025.04 | — | 0.939 | 0 | |
| CLIPBackbone=CLIP-ViT-B/16, Epsilon (ε)=4.02025.04 | — | 0.94 | 0 | |
| EnsembleBackbone=CLIP-ViT-B/16, Epsilon (ε)=4.02025.04 | — | 0.919 | 0.747 | |
| MTABackbone=CLIP-ViT-B/16, Epsilon (ε)=4.02025.04 | — | 0.943 | 0.721 | |
| R-TPTBackbone=CLIP-ViT-B/16, Epsilon (ε)=4.02025.04 | — | 0.937 | 0.82 | |
| TPTBackbone=CLIP-ViT-B/16, Epsilon (ε)=4.02025.04 | — | 0.941 | 0 |