Research map / 3D vision / Depth and stereo
OmniPoint: Universal Monocular Metric Pointcloud from Any Camera
Botao Ye, Marc Pollefeys, Ming-Hsuan Yang, Abhijit Kundu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| predict ray direction plus radial distance instead of XYZ (ablation) | OmniPoint · 360 images | Rel: 6.28 -> 5.79 | Improved |
| predict ray direction plus radial distance instead of XYZ (ablation) | OmniPoint · small FoV | Rel: 4.28 -> 4.04 | Improved |
| predict ray direction plus radial distance instead of XYZ (ablation) | OmniPoint · large FoV | Rel: 6.74 -> 6.37 | Improved |
| use ground-truth-ray point loss instead of naive entangled point loss (ablation) | OmniPoint · point Rel | Rel: 5.65 -> 5.55 | Improved |
| add ground-truth-ray point loss to a disjoint distance and ray loss (ablation) | OmniPoint · point Rel | Rel: 7.88 -> 5.55 | Improved |
| add Perspective-to-Any synthesis (ablation) | OmniPoint · 360 images | Rel: 8.45 -> 6.21 | Improved |
| add Any-to-Perspective self-training alone (ablation) | OmniPoint · 360 images | Rel: 8.45 -> 8.21 | Improved |
| combine both augmentations (ablation) | OmniPoint · 360 images | Rel: 6.21 -> 5.79 | Improved |
| add Gaussian smoothing of sparse depth (ablation) | OmniPoint with sparse depth · relative depth | Rel: 3.80 -> 3.21 | Improved |
| add learnable input-state embeddings (ablation) | OmniPoint with sparse depth · relative depth | Rel: 3.38 -> 3.21 | Improved |
| use explicit ray-distance instead of spherical harmonics ray field (ablation) | OmniPoint · 360 images | Rel: 6.01 -> 5.79 | Improved |
| use explicit ray-distance instead of spherical harmonics ray field (ablation) | OmniPoint · small FoV | Rel: 4.37 -> 4.04 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- UniK3D — UniK3D: Universal Camera Monocular 3D Estimation (CVPR 2025) · better on 5 of 5 shared results
- MoGe V2 — MoGe: Unlocking Accurate Monocular Geometry Estimation for Open-Domain Images with Optimal Training Supervision (CVPR 2025) · better on 2 of 3 shared results
- Depth Pro — Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- UniDepth: Universal Monocular Metric Depth Estimation (CVPR 2024 · similarity 0.90)
- UniK3D: Universal Camera Monocular 3D Estimation (CVPR 2025 · similarity 0.89)
- MoGe-2: Accurate Monocular Geometry with Metric Scale and Sharp Details (NeurIPS 2025 · similarity 0.87)
- Fine-Detail Monocular Geometry Estimation with Self-Guided Sparse Volumetric Refinement (NeurIPS 2026 · similarity 0.87)
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.87)
- MoGe: Unlocking Accurate Monocular Geometry Estimation for Open-Domain Images with Optimal Training Supervision (CVPR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
