Clustering on Food101 (test)
92.8AccuracyTURTLE
Evaluation Results
| Method | Links | |
|---|---|---|
| TURTLEBackbone=Concatenated CLIP ViT-L/14 + DINOv2 ViT-g/142024.06 | 92.8 | |
| HUME CLIPBackbone=CLIP ViT-L/142024.06 | 86.5 | |
| HUME DINOv2Backbone=DINOv2 ViT-g/142024.06 | 85.3 | |
| K-MeansBackbone=Concatenated CLIP ViT-L/14 + DINOv2 ViT-g/142024.06 | 78 |