Depth Estimation on NYU v2 (RMSE, Delta 1-3, REL)
0.22RMSEMREP
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| MREPmrep_component=enabled2026.07 | 0.22 | 97.8 | 99.7 | 99.9 | 0.06 | |
| TADP2026.07 | 0.225 | 97.6 | 99.7 | 99.9 | 0.062 | |
| VPD(LS)variant=LS2026.07 | 0.235 | 97.1 | 99.6 | 99.9 | 0.064 | |
| VPD(R)variant=R2026.07 | 0.248 | 96.5 | 99.5 | 99.9 | 0.068 | |
| VPD2026.07 | 0.254 | 96.4 | 99.5 | 99.9 | 0.069 | |
| w/o MREPmrep_component=disabled2026.07 | 0.256 | 95.1 | 99.4 | 99.9 | 0.071 | |
| ZoeDepth2026.07 | 0.27 | 95.5 | 99.5 | 99.9 | 0.075 | |
| AiT2026.07 | 0.275 | 95.4 | 99.4 | 99.9 | 0.076 | |
| SwinV2-L2026.07 | 0.287 | 94.9 | 99.4 | 99.9 | 0.083 | |
| LingBot-Vision ViT-gNumber of Parameters=1B, Patch Size=16, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.296 | — | — | — | — | |
| V-JEPA 2.1 ViT-GNumber of Parameters=2B, Patch Size=16, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.307 | — | — | — | — | |
| DINOv3Number of Parameters=7B, Patch Size=16, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.309 | — | — | — | — | |
| AM-RADIOv2.5Number of Parameters=1B, Patch Size=14, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.34 | — | — | — | — | |
| V-JEPA 2.1 ViT-gNumber of Parameters=1B, Patch Size=16, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.35 | — | — | — | — | |
| DINOv3 ViT-H+Number of Parameters=0.8B, Patch Size=16, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.352 | — | — | — | — | |
| PEspatialNumber of Parameters=2B, Patch Size=14, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.362 | — | — | — | — | |
| DINOv2Number of Parameters=1B, Patch Size=14, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.372 | — | — | — | — | |
| Web-DINONumber of Parameters=7B, Patch Size=14, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.466 | — | — | — | — | |
| InternVideo2-1BNumber of Parameters=1B, Patch Size=14, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.471 | — | — | — | — | |
| SigLIP 2Number of Parameters=1B, Patch Size=16, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.494 | — | — | — | — | |
| PEcoreNumber of Parameters=2B, Patch Size=14, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.59 | — | — | — | — | |
| V-JEPA2 ViT-gNumber of Parameters=1B, Patch Size=16, Evaluation Protocol=Linear probing on frozen features, Evaluation Resolution=default2026.07 | 0.642 | — | — | — | — | |
| DepthAnythingMethod category=Geometry-specialized2026.07 | — | 98.1 | — | — | 4.3 | |
| DepthAnything V2Method category=Geometry-specialized2026.07 | — | 97.9 | — | — | 4.5 | |
| DICEPTIONMethod category=Generation-based2026.07 | — | 96 | — | — | 6.1 | |
| FE2EMethod category=Generation-based2026.07 | — | 97.7 | — | — | 4.1 | |
| Lotus-2Method category=Generation-based2026.07 | — | 97.6 | — | — | 4.1 | |
| MarigoldMethod category=Generation-based2026.07 | — | 96.4 | — | — | 5.5 | |
| MoGe-2Method category=Geometry-specialized2026.07 | — | 98 | — | — | 3.5 | |
| SenseNova-Visiondepth evaluation protocol=affine-invariant2026.07 | — | 98.1 | — | — | — | |
| SenseNova-Visionprotocol=affine-invariant2026.07 | — | 98.1 | — | — | — | |
| SenseNova-VisionMethod category=Generation-based2026.07 | — | 98.1 | — | — | 4 | |
| Vision Bananadepth evaluation protocol=absolute-depth2026.07 | — | 94.8 | — | — | — | |
| Youtu-VL2026.07 | — | 90.4 | — | — | — |