Object Detection on OV-LVIS v1.0 (test)
34.9mAPrF-ViT+CLIPSelf
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| F-ViT+CLIPSelfBackbone=ViT-L/14, Feature Source=Image Patch2023.10 | 34.9 | 34.6 | 35.6 | 35.1 | |
| RO-ViTBackbone=ViT-H/162023.10 | 34.1 | — | — | 35.1 | |
| CFM-ViTBackbone=ViT-L/162023.10 | 33.9 | — | — | 36.6 | |
| DeticBackbone=SwinB2023.10 | 33.8 | — | — | 47 | |
| F-VLMBackbone=RN50x642023.10 | 32.8 | — | — | 34.9 | |
| RO-ViTBackbone=ViT-L/162023.10 | 32.1 | — | — | 34 | |
| F-VLMBackbone=RN50x162023.10 | 30.4 | — | — | 32.1 | |
| CORA+Backbone=RN50x4, learnable category prompt=true, open-vocabulary ability from additional image annotations=true2023.10 | 28.1 | — | — | — | |
| RO-ViTBackbone=ViT-B/162023.10 | 28 | — | — | 30.2 | |
| VLDetBackbone=SwinB2023.10 | 26.3 | 39.4 | 41.9 | 38.1 | |
| F-VLMBackbone=RN50x42023.10 | 26.3 | — | — | 28.5 | |
| OWL-ViTBackbone=ViT-L/142023.10 | 25.6 | — | — | 34.7 | |
| F-ViT+CLIPSelfBackbone=ViT-B/16, Feature Source=Image Patch2023.10 | 25.3 | 21.8 | 29.1 | 25.2 | |
| DeticBackbone=RN502023.10 | 24.9 | — | — | 32.4 | |
| F-ViTBackbone=ViT-L/142023.10 | 24.2 | 27.9 | 31.5 | 28.7 | |
| BARON-KDBackbone=RN50, learnable category prompt=true2023.10 | 22.6 | 27.6 | 29.8 | 27.6 | |
| CORABackbone=RN50x4, learnable category prompt=true2023.10 | 22.2 | — | — | — | |
| RegionCLIPBackbone=RN50x42023.10 | 22 | 32.1 | 36.9 | 32.3 | |
| VLDetBackbone=RN502023.10 | 21.7 | 29.8 | 34.3 | 30.1 | |
| OADPBackbone=RN502023.10 | 21.7 | 26.3 | 29 | 26.6 | |
| OC-OVDBackbone=RN50, open-vocabulary ability from additional image annotations=true2023.10 | 21.1 | 25 | 29.1 | 25.9 | |
| DetProBackbone=RN50, learnable category prompt=true2023.10 | 19.8 | 25.6 | 28.9 | 25.9 | |
| F-VLMBackbone=RN502023.10 | 18.6 | — | — | 24.2 | |
| OV-DETRBackbone=RN502023.10 | 17.4 | 25 | 32.5 | 26.6 | |
| RegionCLIPBackbone=RN502023.10 | 17.1 | 27.4 | 34 | 28.2 | |
| ViLDBackbone=RN502023.10 | 16.6 | 24.6 | 30.3 | 25.5 | |
| F-ViTBackbone=ViT-B/162023.10 | 11.5 | 12.3 | 20.6 | 15.4 |