Research map / 3D vision / Depth and stereo
Any Resolution Any Geometry: From Multi-View To Multi-Patch
Wenqing Cui, Zhenyu Li, Mykola Lavreniuk, Jian Shi, Ramzi Idoughi, Xiangjun Tang, Peter Wonka
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 GridMix mixed-grid sampling instead of single 1x1 grid (ablation) | URGT · UnrealStereo4K depth | AbsRel: 0.0500 -> 0.0295 | Improved |
| add GridMix mixed-grid sampling instead of fixed 4x4 grid (ablation) | URGT · UnrealStereo4K depth | CE: 0.0635 -> 0.0418 | Improved |
| add GridMix mixed-grid sampling instead of fixed 4x4 grid (ablation) | URGT · UnrealStereo4K depth | AbsRel: 0.0321 -> 0.0295 | Improved |
| use global RoPE instead of local RoPE (ablation) | URGT · UnrealStereo4K depth | CE: 0.2830 -> 0.0635 | Improved |
| use global RoPE instead of local RoPE (ablation) | URGT · UnrealStereo4K depth | AbsRel: 0.0343 -> 0.0321 | Improved |
| add cross-patch attention (ablation) | URGT · UnrealStereo4K depth | AbsRel: 0.0678 -> 0.0500 | Improved |
| add cross-patch attention (ablation) | URGT · UnrealStereo4K depth | RMSE: 2.51 -> 2.01 | Improved |
| train jointly with normals instead of depth only (ablation) | URGT · UnrealStereo4K depth | AbsRel: 0.0295 -> 0.0291 | Improved |
| train jointly with normals instead of depth only (ablation) | URGT · Middlebury 2014 zero-shot depth | AbsRel: 0.0291 -> 0.0280 | Improved |
| use URGT multi-patch transformer instead of PatchRefiner p=49 | URGT · UnrealStereo4K depth | AbsRel: 0.0582 -> 0.0291 | Improved |
| use URGT instead of Metric3D v2 for normals | URGT · UnrealStereo4K normal | Mean angular error: 23.36 -> 18.51 | Improved |
| use URGT instead of Depth-Anything v2 on Booster | URGT · Booster zero-shot depth | AbsRel: 0.0274 -> 0.0248 | Improved |
Closest papers by meaning
- One Look is Enough: Seamless Patchwise Refinement for Zero-Shot Monocular Depth Estimation on High-Resolution Images (ICCV 2025 · similarity 0.88)
- PatchFusion: An End-to-End Tile-Based Framework for High-Resolution Monocular Metric Depth Estimation (CVPR 2024 · similarity 0.88)
- PatchRefiner: Leveraging Synthetic Data for Real-Domain High-Resolution Monocular Metric Depth Estimation (ECCV 2024 · similarity 0.87)
- RegimeVGGT: Layer-Wise Spatially Preserving Redundancy Removal for Visual Geometry Grounded Transformer (NeurIPS 2026 · similarity 0.86)
- Ultra-High Resolution Segmentation via Boundary-Enhanced Patch-Merging Transformer (AAAI 2025 · similarity 0.86)
- Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025 · similarity 0.85)
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.
