Image Classification on 11 datasets average CLIP ResNet-101 (test)
82AccDANCE
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DANCEBackbone=CLIP ResNet-101, alpha=0.12026.02 | 82 | 4.11 | 92.6 | — | 8.746 | |
| DANCEBackbone=CLIP ResNet-101, alpha=0.052026.02 | 82 | 8.07 | 96.2 | — | 5.627 | |
| k-NN SetBackbone=CLIP ResNet-101, alpha=0.12026.02 | 82 | 4.03 | 92 | — | 9.006 | |
| k-NN SetBackbone=CLIP ResNet-101, alpha=0.052026.02 | 82 | 8.47 | 95.7 | — | 5.963 | |
| CLR SetBackbone=CLIP ResNet-101, alpha=0.12026.02 | 82 | 7.49 | 90.3 | — | 8.626 | |
| CLR SetBackbone=CLIP ResNet-101, alpha=0.052026.02 | 82 | 9.96 | 96.5 | — | 5.43 | |
| Deep k-NNBackbone=CLIP ResNet-101, alpha=0.12026.02 | 82 | 5.66 | 93.5 | — | 8.855 | |
| Deep k-NNBackbone=CLIP ResNet-101, alpha=0.052026.02 | 82 | 11.58 | 97.2 | — | 5.018 | |
| RFM Adapter (APS)Backbone=CLIP ResNet-101, alpha=0.12026.02 | 82 | 3.13 | 90.4 | — | 8.761 | |
| RFM Adapter (APS)Backbone=CLIP ResNet-101, alpha=0.052026.02 | 82 | 8.96 | 95.4 | — | 5.982 | |
| RFM Adapter (RAPS)Backbone=CLIP ResNet-101, alpha=0.12026.02 | 82 | 3.12 | 90.3 | — | 8.796 | |
| RFM Adapter (RAPS)Backbone=CLIP ResNet-101, alpha=0.052026.02 | 82 | 8.98 | 95.4 | — | 6.002 | |
| NCP (RAPS)Backbone=CLIP ResNet-101, alpha=0.12026.02 | 82 | 3.38 | 92.6 | — | 8.904 | |
| NCP (RAPS)Backbone=CLIP ResNet-101, alpha=0.052026.02 | 82 | 8.85 | 96 | — | 5.66 | |
| Conf-OT (APS)Backbone=CLIP ResNet-101, alpha=0.12026.02 | 66.1 | 8.44 | 90.7 | — | 8.604 | |
| Conf-OT (APS)Backbone=CLIP ResNet-101, alpha=0.052026.02 | 66.1 | 13.95 | 95.4 | — | 5.82 | |
| Conf-OT (RAPS)Backbone=CLIP ResNet-101, alpha=0.12026.02 | 66.1 | 7.39 | 90.7 | — | 8.651 | |
| Conf-OT (RAPS)Backbone=CLIP ResNet-101, alpha=0.052026.02 | 66.1 | 11.57 | 95.5 | — | 5.859 |