| Paper | Code | % < 11.25 | % < 22.5 | % < 30 | Mean Angle Error | RMSE | ModelName | ReleaseDate |
|---|---|---|---|---|---|---|---|---|
| Metric3Dv2: A Versatile Monocular Geometric Foundation Model for Zero-shot Metric Depth and Surface Normal Estimation | ✓ Link | 68.8 | 84.9 | 89.8 | 12.0 | 19.2 | Metric3Dv2(L, FT) | 2024-03-22 |
| PolyMaX: General Dense Prediction with Mask Transformer | ✓ Link | 65.66 | 82.28 | 87.83 | 13.09 | 20.4 | PolyMaX(ConvNeXt-L) | 2023-11-09 |
| iDisc: Internal Discretization for Monocular Depth Estimation | ✓ Link | 63.8 | 79.8 | 85.6 | 14.6 | 22.8 | iDisc | 2023-04-13 |
| Estimating and Exploiting the Aleatoric Uncertainty in Surface Normal Estimation | ✓ Link | 62.2 | 79.3 | 85.2 | 14.9 | 23.5 | Bae et al. | 2021-09-20 |
| Fine-Tuning Image-Conditional Diffusion Models is Easier than You Think | ✓ Link | 61.4 | 16.2 | Marigold + E2E FT(zero-shot) | 2024-09-17 | |||
| Floors are Flat: Leveraging Semantics for Real-Time Surface Normal Prediction | ✓ Link | 59.5 | 72.2 | 77.3 | 19.7 | 19.3 | Floors are Flat | 2019-06-16 |