Object Detection on COCO-O (test)
53.7Average PerformanceFrozen-DETR (DINO-det+CLIP+DINOv2)
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Frozen-DETR (DINO-det+CLIP+DINOv2)Backbone=R502024.10 | 53.7 | 52.8 | 49.3 | 53.5 | 56.6 | 57.7 | 52.3 | 29.49 | |
| Frozen-DETR (DINO-det+CLIP)Backbone=R502024.10 | 50.2 | 50.3 | 46.3 | 51.4 | 54.9 | 52.1 | 46 | 26.8 | |
| DINO-det-5scaleBackbone=Swin-L2024.10 | 42.1 | — | — | — | — | — | — | 15.76 | |
| ViTDetBackbone=ViT-H2024.10 | 34.3 | — | — | — | — | — | — | 7.89 | |
| DINO-det-4scaleBackbone=R502024.10 | 22.5 | 13.8 | 36.3 | 18.5 | 30.7 | 13.3 | 22.4 | 0.45 | |
| Deformable DETRBackbone=R502024.10 | 18.5 | 10.5 | 30.2 | 15.1 | 26.2 | 10.6 | 18.6 | -1.49 | |
| DETRBackbone=R502024.10 | 17.1 | 9 | 30 | 12.3 | 23.9 | 11.6 | 15.7 | -1.82 |