Research map / 3D vision / Depth and stereo
FoundationGeo: Learning Spatial Pixel-Wise Fields for Monocular Metric Geometry
Muxin Liu, Xiaoyang Lyu, Tianhe Ren, Peng Dai, Xiaoshan Wu, Zhiyue Zhang, Jiaqi Zhang, Jiehong Lin, Shaoshuai Shi, Qi Xiaojuan
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 |
|---|---|---|---|
| add spatial calibration fields instead of direct point-map fine-tuning (ablation) | two-stage FoundationGeo · 7-dataset avg (5% subsets) | AbsRel / delta1: 19.6 -> 19.3 AbsRel; 65.9 -> 68.0 delta1 | Improved |
| replace global scale with pixel-wise spatial fields (ablation) | two-stage FoundationGeo · 7-dataset avg (5% subsets) | AbsRel / delta1: 20.1 -> 19.3 AbsRel; 66.8 -> 68.0 delta1 | Improved |
| add field-specific scale and ray losses (ablation) | FoundationGeo spatial fields · 7-dataset avg (5% subsets) | AbsRel / delta1: 19.3 -> 18.8 AbsRel; 68.0 -> 69.7 delta1 | Improved |
| use two-stage training instead of single-stage (ablation) | FoundationGeo · 7-dataset avg (5% subsets) | AbsRel / delta1: 22.4 -> 19.6 AbsRel; 57.7 -> 65.9 delta1 | Improved |
| initialize from Stage-1 relative base model instead of training from scratch (ablation) | FoundationGeo · 7-dataset avg (full data) | AbsRel / delta1: 15.2/78.2 -> 14.8/80.8 | Improved |
| add diverse-focal rendered FGD data (ablation) | FoundationGeo · 7-dataset avg (full data) | AbsRel / delta1: 15.0 -> 14.8 AbsRel; 79.5 -> 80.8 delta1 | Improved |
| add fixed-focal (1320 px) rendered data (ablation) | FoundationGeo · 7-dataset avg (full data) | AbsRel / delta1: 15.0 -> 15.3 AbsRel; 79.5 -> 78.9 delta1 | Hurt |
| add fixed-focal rendered data (ablation) | FoundationGeo · DDAD | AbsRel: 20.0 -> 19.2 | Improved |
| fine-tune MoGe-2 with FGD rendered data | MoGe-2 · 7-dataset avg | AbsRel / delta1: 15.7 -> 15.1 AbsRel; 76.8 -> 79.0 delta1 | Improved |
| fine-tune MoGe-2 with FGD rendered data | MoGe-2 · DIODE | delta1: 66.4 -> 72.1 | Improved |
| add ray-direction correction (ablation) | FoundationGeo · 4-dataset avg (ray metrics) | MaeDeg: 2.889 -> 2.355 | Improved |
| add ray-direction correction (ablation) | FoundationGeo · 4-dataset avg | delta1: 72.8 -> 76.8 | 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.
- MoGe-2 — MoGe-2: Accurate Monocular Geometry with Metric Scale and Sharp Details (NeurIPS 2025) · better on 5 of 5 shared results
- DepthPro — Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025) · better on 2 of 2 shared results
- MoGe-1 — MoGe: Unlocking Accurate Monocular Geometry Estimation for Open-Domain Images with Optimal Training Supervision (CVPR 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
- MoGe-2: Accurate Monocular Geometry with Metric Scale and Sharp Details (NeurIPS 2025 · similarity 0.89)
- MoGe: Unlocking Accurate Monocular Geometry Estimation for Open-Domain Images with Optimal Training Supervision (CVPR 2025 · similarity 0.88)
- TALO: Pushing 3D Vision Foundation Models Towards Globally Consistent Online Reconstruction (CVPR 2026 · similarity 0.88)
- MetricAnything: Scaling Metric Depth Pretraining with Noisy Heterogeneous Sources (ECCV 2026 · similarity 0.87)
- Feat2GS: Probing Visual Foundation Models with Gaussian Splatting (CVPR 2025 · similarity 0.87)
- SurGe: Improved Surface Geometry in Point Maps (NeurIPS 2026 · similarity 0.86)
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.
