Object Detection on FAIR1M 1.0
41.66mAP50BHCL
Evaluation results
| Method | Links | |
|---|---|---|
| BHCLBackbone=ResNet-50, Input Resolution=1024x10242025.12 | 41.66 | |
| OrientedFormerBackbone=ResNet-50, Input Resolution=1024x10242025.12 | 41.31 | |
| DRNetBackbone=ResNet-50, Input Resolution=1024x10242025.12 | 40.87 | |
| SFRNetBackbone=ResNet-50, Input Resolution=1024x10242025.12 | 40.74 | |
| PETDetBackbone=ResNet-50, Input Resolution=1024x10242025.12 | 40.55 | |
| PCLDetBackbone=ResNet-50, Input Resolution=1024x10242025.12 | 40.39 | |
| RoI Trans. + COBB-ln-sigbackbone=RoI Trans., variant=ln-sig2024.02 | 39.82 | |
| RoI Trans. + COBB-ln-lnbackbone=RoI Trans., variant=ln-ln2024.02 | 39.66 | |
| RoI Trans. + COBB-sig-sigbackbone=RoI Trans., variant=sig-sig2024.02 | 39.61 | |
| RoI Trans.baseline=true2024.02 | 39.31 | |
| ORCNNBackbone=ResNet-50, Input Resolution=1024x10242025.12 | 38.85 | |
| Oriented R-CNN + COBB-lnbackbone=Oriented R-CNN, variant=ln2024.02 | 38.83 | |
| Oriented R-CNN + COBB-sigbackbone=Oriented R-CNN, variant=sig2024.02 | 38.81 | |
| RoITransBackbone=ResNet-50, Input Resolution=1024x10242025.12 | 38.49 | |
| Oriented R-CNNbaseline=true2024.02 | 38.3 | |
| FRCNNBackbone=ResNet-50, Input Resolution=1024x10242025.12 | 36.83 | |
| Rotated Faster R-CNN + COBB-sigbackbone=Rotated Faster R-CNN, variant=sig2024.02 | 36.81 | |
| Rotated Faster R-CNN + COBB-lnbackbone=Rotated Faster R-CNN, variant=ln2024.02 | 36.53 | |
| Gliding Vertexbaseline=true2024.02 | 36.37 | |
| Rotated Faster R-CNNbaseline=true2024.02 | 35.16 |