Object Detection on DOTA 2.0
57.17mAPGGHL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GGHLSpeed (fps)=41.072021.09 | 57.17 | — | |
| Faster R-CNN OBB + RTSpeed (fps)=12.402021.09 | 52.81 | — | |
| Hybrid Task MaskSpeed (fps)=7.902021.09 | 50.34 | — | |
| Cascade Mask R-CNNSpeed (fps)=7.202021.09 | 50.04 | — | |
| Mask R-CNNSpeed (fps)=9.702021.09 | 49.47 | — | |
| Faster R-CNN H-OBBSpeed (fps)=13.702021.09 | 48.9 | — | |
| Faster R-CNN OBB + DpoolSpeed (fps)=12.102021.09 | 48.77 | — | |
| RS-MPODPretrain Dataset=LAE-1M, Prompt=multimodal2026.02 | 48 | 74.6 | |
| DisDopPre-Training Data=LAE-1M, Evaluation Mode=Open-set detection2026.05 | 47.5 | — | |
| Faster R-CNN OBBSpeed (fps)=14.102021.09 | 47.31 | — | |
| LAE-DINOPretrain Dataset=LAE-1M, Prompt=text2026.02 | 46.8 | — | |
| LAE-DINOPre-Training Data=LAE-1M, Evaluation Mode=Open-set detection2026.05 | 46.8 | — | |
| RetinaNet OBBSpeed (fps)=12.102021.09 | 46.68 | — | |
| RS-MPODPretrain Dataset=LAE-1M, Prompt=text2026.02 | 46.1 | 73.8 | |
| GroundingDINOPretrain Dataset=LAE-1M, Prompt=text2026.02 | 46 | — | |
| GroundingDINO with DVCPre-Training Data=LAE-1M, Evaluation Mode=Open-set detection2026.05 | 46 | — | |
| RS-MPODPretrain Dataset=LAE-1M, Prompt=visual–322026.02 | 44.1 | 70.1 | |
| GLIPPretrain Dataset=LAE-1M, Prompt=text2026.02 | 43 | — | |
| GLIP with DVCPre-Training Data=LAE-1M, Evaluation Mode=Open-set detection2026.05 | 43 | — | |
| OpenRSDPretrain Dataset=ORSD+, Prompt=text2026.02 | — | 71.8 | |
| OpenRSDPretrain Dataset=ORSD+, Prompt=visual2026.02 | — | 69.8 |