Multi-view Depth Estimation on Tanks and Temples (T&T)
2Relative Error (rel)PE3R
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PE3RMethod category=semantic reconstruction, Base predictor=VGGT2025.03 | 2 | 85.4 | |
| VGGTMethod category=feed-forward methods, Implementation source=official2025.03 | 2.1 | 85.2 | |
| PE3RMethod category=semantic reconstruction, Base predictor=DUSt3R2025.03 | 2.1 | 85.3 | |
| COLMAPMethod category=classical methods, Implementation source=official2025.03 | 2.7 | 95 | |
| DUSt3RMethod category=feed-forward methods, Implementation source=our implementation2025.03 | 2.9 | 78.5 | |
| DUSt3RResolution=5122025.01 | 3.17 | 76.68 | |
| DUSt3RMethod category=feed-forward methods, Implementation source=official2025.03 | 3.2 | 76.7 | |
| Fast3RMethod category=feed-forward methods, Implementation source=official2025.03 | 4.4 | 64 | |
| Fast3R2025.01 | 4.43 | 63.95 | |
| DUSt3RResolution=2242025.01 | 5.54 | 56.38 | |
| MASt3RMethod category=feed-forward methods, Implementation source=our implementation2025.03 | 22.1 | 14.6 | |
| COLMAP-DENSE2025.01 | 25.7 | 76.4 | |
| COLMAP DenseMethod category=classical methods, Implementation source=official2025.03 | 25.7 | 76.4 |