Depth Estimation on 7-Scenes (test)
0.1Abs RelOurs(+pose,atten)
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Ours(+pose,atten)training_dataset=ScanNet, generalization=zero-shot, sampling_rate=every 10 frames, frames_per_sample=52022.05 | 0.1 | 0.1781 | 0.2664 | — | 89.67 | 0.0473 | |
| Ours(base)training_dataset=ScanNet, generalization=zero-shot, sampling_rate=every 10 frames, frames_per_sample=52022.05 | 0.1148 | 0.1999 | 0.2857 | — | 87.26 | 0.0552 | |
| PairNettraining_dataset=ScanNet, generalization=zero-shot, sampling_rate=every 10 frames, frames_per_sample=52022.05 | 0.1157 | 0.2086 | 0.2926 | — | 87.68 | 0.0677 | |
| MaGNet (full)Cap=10m2021.12 | 0.1257 | 0.2133 | 0.2957 | 0.1639 | 85.52 | — | |
| MaGNet (full)Cap=5m2021.12 | 0.1257 | 0.2133 | 0.2957 | 0.1639 | 85.52 | — | |
| IterMVStraining_dataset=ScanNet, generalization=zero-shot, sampling_rate=every 10 frames, frames_per_sample=52022.05 | 0.1336 | 0.2363 | 0.3425 | — | 85.18 | 0.1033 | |
| MaGNet (D-Net)Cap=10m2021.12 | 0.1339 | 0.2209 | 0.2932 | 0.1677 | 83.08 | — | |
| MaGNet (D-Net)Cap=5m2021.12 | 0.1339 | 0.2209 | 0.2932 | 0.1677 | 83.08 | — | |
| Long et al.Cap=10m2021.12 | 0.1465 | 0.2528 | 0.3382 | 0.1967 | 80.36 | — | |
| Long et al.Cap=5m2021.12 | 0.1465 | 0.2528 | 0.3382 | 0.1967 | 80.36 | — | |
| ESTDtraining_dataset=ScanNet, generalization=zero-shot, sampling_rate=every 10 frames, frames_per_sample=52022.05 | 0.1465 | 0.2528 | 0.3382 | — | 80.36 | 0.0729 | |
| DELTASCap=10m2021.12 | 0.1548 | 0.2671 | 0.3541 | 0.186 | 79.66 | — | |
| CNM-NetCap=10m2021.12 | 0.1602 | 0.2751 | 0.3602 | 0.203 | 76.81 | — | |
| NASCap=10m2021.12 | 0.1631 | 0.2885 | 0.3791 | 0.1997 | 77.12 | — | |
| DPSNetCap=10m2021.12 | 0.1675 | 0.297 | 0.3905 | 0.2061 | 76.03 | — | |
| MVDepthNetCap=10m2021.12 | 0.1905 | 0.3304 | 0.426 | 0.2221 | 71.93 | — | |
| UCS-NetCap=10m2021.12 | 0.2113 | 0.3668 | 0.4683 | 0.2369 | 69.31 | — | |
| NeuralRGBDCap=5m2021.12 | 0.2334 | 0.406 | 0.5358 | 0.2516 | 68.03 | — | |
| NRGBDtraining_dataset=ScanNet, generalization=zero-shot, sampling_rate=every 10 frames, frames_per_sample=52022.05 | 0.2334 | 0.406 | 0.5358 | — | 68.03 | 0.2163 |