Monocular Depth Estimation on Average 7 Datasets
0.969Delta 1 AccMoGe
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MoGeFine-tuning=false, Evaluation Protocol=Scale-invariant relative depth2025.11 | 0.969 | 0.0414 | |
| VGGT+OursFine-tuning=true, Evaluation Protocol=Scale-invariant relative depth2025.11 | 0.967 | 0.0429 | |
| DUSt3R+OursFine-tuning=true, Evaluation Protocol=Scale-invariant relative depth2025.11 | 0.948 | 0.0558 | |
| CUT3R+OursFine-tuning=true, Evaluation Protocol=Scale-invariant relative depth2025.11 | 0.947 | 0.0559 | |
| VGGTFine-tuning=false, Evaluation Protocol=Scale-invariant relative depth2025.11 | 0.94 | 0.0577 | |
| CUT3RFine-tuning=false, Evaluation Protocol=Scale-invariant relative depth2025.11 | 0.929 | 0.0646 | |
| DUSt3RFine-tuning=false, Evaluation Protocol=Scale-invariant relative depth2025.11 | 0.923 | 0.0703 | |
| Vision BananaCamera Intrinsics (Inference)=No, Camera Intrinsics (Training)=No, Zero-shot transfer=Yes2026.04 | 0.882 | 0.116 | |
| UniK3DCamera Intrinsics (Inference)=No, Camera Intrinsics (Training)=Yes, Zero-shot transfer=Yes2026.04 | 0.823 | 0.156 | |
| MoGe-2Camera Intrinsics (Inference)=No, Camera Intrinsics (Training)=Yes, Zero-shot transfer=Yes2026.04 | 0.802 | 0.144 | |
| MoGe-2Fine-tuning=false, Evaluation Protocol=Metric depth2025.11 | 0.786 | 0.1482 | |
| Depth ProCamera Intrinsics (Inference)=No, Camera Intrinsics (Training)=Yes, Zero-shot transfer=Yes2026.04 | 0.715 | — | |
| MASt3R+OursFine-tuning=true, Evaluation Protocol=Metric depth2025.11 | 0.643 | 0.276 | |
| MASt3RFine-tuning=false, Evaluation Protocol=Metric depth2025.11 | 0.302 | 0.4962 |