Multi-object Category Recognition (Categ-MObj) on Toys4k multi-object setting
60.49LSAImageBind
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ImageBindBackbone=ViT H/162023.12 | 60.49 | — | |
| DINOv2Backbone=ViT B/142023.12 | 57.92 | — | |
| DINOv1Backbone=ViT S/82023.12 | 56.99 | — | |
| DINOv2Backbone=ViT S/142023.12 | 56.95 | — | |
| CLIP-ImgBackbone=ViT B/162023.12 | 56.76 | — | |
| DINOv2Backbone=ViT B/142023.12 | 39.24 | 50.88 | |
| ImageBindBackbone=ViT H/162023.12 | 38.85 | 50.24 | |
| CLIP-ImgBackbone=ViT B/162023.12 | 38.25 | 48.96 | |
| DINOv2Backbone=ViT S/142023.12 | 37.08 | 48.16 | |
| DINOv1Backbone=ViT S/82023.12 | 36.44 | 46.92 |