Table 3.
OD and OC segmentation results on Drishti-GS and REFUGE datasets.
| Datasets | Methods | OD Segmentation | OC Segmentation | ||
|---|---|---|---|---|---|
| DC | JAC | DC | JAC | ||
| REFUGE | M-Net [12] | 0.943 | - | 0.831 | - |
| M-Ada [25] | 0.958 | - | 0.882 | - | |
| EARDS [27] | 0.954 | 0.914 | 0.887 | 0.801 | |
| pOSAL [48] | 0.946 | - | 0.875 | - | |
| Multi-Model [49] | - | 0.922 | - | 0.790 | |
| CFEA [50] | 0.941 | - | 0.862 | - | |
| Two-Stage Mask R-CNN [51] | 0.947 | - | 0.854 | - | |
| Ours | 0.965 | 0.933 | 0.902 | 0.824 | |
| ORIGA | Deep object detection Network [7] |
0.845 | - | 0.845 | - |
| JointRCNN [19] | 0.937 | - | 0.794 | - | |
| SS-DCGAN [38] | 0.901 | - | - | - | |
| Ours | 0.962 | 0.928 | 0.871 | 0.773 | |
| Drishti-GS | U-Net [3] | 0.950 | - | 0.800 | - |
| [11] | 0.973 | 0.949 | 0.887 | 0.804 | |
| FC-DenseNet [12] | 0.949 | 0.904 | 0.828 | 0.711 | |
| M-Net [14] | 0.959 | - | 0.866 | - | |
| M-Ada [25] | 0.971 | - | 0.910 | - | |
| EARDS [27] | 0.974 | 0.949 | 0.915 | 0.849 | |
| ResFPN-Net [29] | 0.976 | - | 0.896 | - | |
| WRoIM [52] | 0.960 | - | 0.890 | - | |
| WGAN [53] | 0.954 | - | 0.840 | - | |
| pOSAL [48] | 0.965 | - | 0.858 | - | |
| GL-Net [54] | 0.971 | - | 0.905 | - | |
| Multi-Model [49] | 0.960 | 0.924 | 0.902 | 0.822 | |
| Ours | 0.943 | 0.893 | 0.889 | 0.801 | |
| RIM-ONE -V3 |
Hybrid [8] | 0.930 | 0.910 | 0.910 | 0.880 |
| Modified U-Net [9] | 0.950 | 0.890 | 0.820 | 0.690 | |
| ECSD [32] | 0.860 | 0.760 | 0.800 | 0.680 | |
| EE-U-Net [34] | 0.950 | 0.880 | 0.860 | 0.760 | |
| pOSAL [48] | 0.860 | - | 0.787 | - | |
| Ours | 0.910 | 0.830 | 0.640 | 0.770 | |