Table 2.
Comparison between rotated detectors and horizontal detectors.
| Mode | Method | AP (%) |
|---|---|---|
| Horizontal detection methods | Autoassign (Zhu et al., 2020) | 63.5 |
| Fcos (Tian et al., 2019) | 65.3 | |
| Ddod (Chen et al., 2021) | 64.9 | |
| Dynamic Head (Dai et al., 2021) | 67.6 | |
| Libra-rcnn (Pang et al., 2019) | 64.7 | |
| Faster-rcnn (Ren et al., 2017) | 68.6 | |
| Cascade-rcnn (Cai and Vasconcelos, 2018) | 69.1 | |
| Doublehead-rcnn (Wu et al., 2020) | 69.3 | |
| Rotated detection methods | Gliding Vertex (Xu et al., 2021) | 66.1 |
| Rotated Fcos (Tian et al., 2019) | 67.2 | |
| Oriented Reppoints (Li et al., 2022) | 70 | |
| S2anet (Han et al., 2021) | 70.7 | |
| Rotated Faster-rcnn (Ren et al., 2017) | 69.4 | |
| Roi-trans (Ding et al., 2019) | 72.4 | |
| CMRD-Net | 73.7 |