Multi-view Depth Estimation on KITTI
0.0353Absolute Relative Error (Rel)Ours
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| OursScale=GT2026.06 | 0.0353 | — | 0.1219 | 2.0039 | 0.0634 | 98.96 | 99.72 | 99.89 | |
| PI3Scale=GT2026.06 | 0.0464 | — | 0.1488 | 2.1142 | 0.0748 | 98.28 | 99.7 | 99.89 | |
| DA3Scale=GT2026.06 | 0.0473 | — | 0.2601 | 2.5869 | 0.0875 | 97.54 | 99.38 | 99.78 | |
| VGGTScale=GT2026.06 | 0.0703 | — | 0.2904 | 3.2424 | 0.1064 | 96.18 | 98.99 | 99.66 | |
| Mast3RScale=GT2026.06 | 0.0834 | — | 0.6435 | 4.6779 | 0.3605 | 94.45 | 97.74 | 98.61 | |
| Fast3RScale=GT2026.06 | 0.1024 | — | 0.5489 | 3.9426 | 0.1501 | 90.53 | 97.89 | 99.35 | |
| OursScale=Pred2026.06 | 0.1226 | — | 0.3605 | 2.946 | 0.1475 | 85.9 | 99.88 | 99.89 | |
| DA3Scale=Pred2026.06 | 0.2448 | — | 1.192 | 5.0979 | 0.3014 | 35.88 | 85.82 | 97.54 | |
| DUSt3RMethod category=feed-forward methods, Implementation source=official2025.03 | 9.1 | 39.5 | — | — | — | — | — | — | |
| PE3RMethod category=semantic reconstruction, Base predictor=VGGT2025.03 | 9.2 | 43.8 | — | — | — | — | — | — | |
| VGGTMethod category=feed-forward methods, Implementation source=official2025.03 | 9.4 | 41.3 | — | — | — | — | — | — | |
| PE3RMethod category=semantic reconstruction, Base predictor=DUSt3R2025.03 | 9.4 | 48.6 | — | — | — | — | — | — | |
| DUSt3RMethod category=feed-forward methods, Implementation source=our implementation2025.03 | 11 | 33.2 | — | — | — | — | — | — | |
| COLMAPMethod category=classical methods, Implementation source=official2025.03 | 12 | 58.2 | — | — | — | — | — | — | |
| COLMAP DenseMethod category=classical methods, Implementation source=official2025.03 | 26.9 | 52.7 | — | — | — | — | — | — | |
| MASt3RMethod category=feed-forward methods, Implementation source=our implementation2025.03 | 36.9 | 5.4 | — | — | — | — | — | — |