Depth estimation on Stanford2D3D (fold-1 test)
0.1014MREHoHoNet
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| HoHoNetBackbone=ResNet-50, Latent size D=256, IDCT components r=642020.11 | 0.1014 | 0.2027 | 0.3834 | 0.0668 | 90.54 | 96.93 | 98.86 | |
| BiFuseInput=ERP and cubemap2020.11 | 0.1209 | 0.2343 | 0.4142 | 0.0787 | 86.6 | 95.8 | 98.6 | |
| Cube2020.11 | 0.1332 | 0.2588 | 0.4407 | 0.0844 | 83.47 | 95.23 | 98.38 | |
| Equi2020.11 | 0.1428 | 0.2711 | 0.4637 | 0.0911 | 82.61 | 94.58 | 98 | |
| FCRN2020.11 | 0.1837 | 0.3428 | 0.5774 | 0.11 | 72.3 | 92.07 | 97.31 | |
| OmniDepth (bn)2020.11 | 0.1996 | 0.3743 | 0.6152 | 0.1212 | 68.77 | 88.91 | 95.78 |