Glass Segmentation on GSD
93.8IoUL+GNet
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| L+GNettraining=all datasets, binary prediction confidences=true2026.03 | 93.8 | 96.5 | 0.022 | 2.28 | |
| GlassWizardtraining=all datasets2026.03 | 90.4 | 95.2 | 0.036 | 4.03 | |
| L+GNettraining mode=trained and tested separately on respective datasets, with binary prediction confidences=true2026.03 | 0.931 | 0.961 | 0.024 | 2.54 | |
| GlassWizardtraining mode=trained and tested separately on respective datasets2026.03 | 0.891 | 0.942 | 0.035 | 4.14 | |
| VBNettraining mode=trained and tested separately on respective datasets2026.03 | 0.861 | 0.921 | 0.043 | 5.51 | |
| GlassSemNettraining mode=trained and tested separately on respective datasets2026.03 | 0.856 | 0.92 | 0.044 | 5.6 | |
| GSDNettraining mode=trained and tested separately on respective datasets2026.03 | 0.836 | 0.903 | 0.055 | 6.12 | |
| PGSNettraining mode=trained and tested separately on respective datasets2026.03 | 0.836 | — | 0.054 | 6.25 | |
| GDNettraining mode=trained and tested separately on respective datasets2026.03 | 0.825 | — | 0.058 | 6.41 | |
| EBLNettraining mode=trained and tested separately on respective datasets2026.03 | 0.817 | 0.878 | 0.059 | 6.75 | |
| Trans2Segtraining mode=trained and tested separately on respective datasets2026.03 | 0.797 | — | 0.069 | 8.21 | |
| TransLabtraining mode=trained and tested separately on respective datasets2026.03 | 0.781 | — | 0.069 | 9.19 | |
| SAMtraining mode=trained and tested separately on respective datasets2026.03 | 0.506 | 0.799 | 0.213 | 23.91 |