Low-shot recognition on Toys4k multi-object setting (test)
60.49LSACateg-MObj
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Categ-MObjBackbone=ImageBind ViT H/162023.12 | 60.49 | — | |
| Categ-MObjBackbone=DINOv2 ViT B/142023.12 | 57.92 | — | |
| Categ-MObjBackbone=DINOv1 ViT S/82023.12 | 56.99 | — | |
| Categ-MObjBackbone=DINOv2 ViT S/142023.12 | 56.95 | — | |
| Categ-MObjBackbone=CLIP-Img ViT B/162023.12 | 56.76 | — | |
| Categ-MObj SuppAssignBackbone=ImageBind ViT H/162023.12 | 45.91 | 58.58 | |
| Categ-MObj SuppAssignBackbone=DINOv2 ViT B/142023.12 | 43.21 | 54.96 | |
| Categ-MObj SuppAssignBackbone=DINOv2 ViT S/142023.12 | 41.26 | 52.28 | |
| Categ-MObj SuppAssignBackbone=CLIP-Img ViT B/162023.12 | 41.22 | 51.64 | |
| Categ-MObj SuppAssignBackbone=DINOv1 ViT S/82023.12 | 40.21 | 51.68 | |
| LSMEBackbone=DINOv2 ViT B/142023.12 | 39.24 | 50.88 | |
| LSMEBackbone=ImageBind ViT H/162023.12 | 38.85 | 50.24 | |
| LSMEBackbone=CLIP-Img ViT B/162023.12 | 38.25 | 48.96 | |
| LSMEBackbone=DINOv2 ViT S/142023.12 | 37.08 | 48.16 | |
| LSMEBackbone=DINOv1 ViT S/82023.12 | 36.44 | 46.92 |