Research map / 3D vision / Depth and stereo
PatchRefiner: Leveraging Synthetic Data for Real-Domain High-Resolution Monocular Metric Depth Estimation
Zhenyu Li, Shariq Farooq Bhat, 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 |
|---|---|---|---|
| reformulate high-resolution depth as residual refinement of a frozen coarse model (PatchRefiner vs PatchFusion) | ZoeDepth · UnrealStereo4K | RMS: 1.088 -> 0.892 | Improved |
| predict residual on coarse depth instead of direct depth (ablation) | PatchRefiner · UnrealStereo4K | RMS: 0.925 -> 0.892 | Improved |
| add more feature levels (1 to 6) to refiner decoder (ablation) | PatchRefiner · UnrealStereo4K | RMS: 0.926 -> 0.892 | Improved |
| add DSD loss with pseudo labels from synthetic teacher | PatchRefiner · CityScapes | Boundary Recall: 40.92 -> 48.78 | Improved |
| mix synthetic and real data during training | PatchRefiner · CityScapes | delta1 (%): 95.418 -> 89.108 | Hurt |
| fine-tune on real data after synthetic pretraining | PatchRefiner · CityScapes | Boundary F1: 23.04 -> 23.68 | No clear effect |
| supervise pseudo labels with silog loss instead of rank+ssi (ablation) | PatchRefiner · CityScapes | delta1 (%): 95.359 -> 82.550 | Hurt |
| restrict silog pseudo-label loss to regions lacking GT (masking) (ablation) | PatchRefiner · CityScapes | delta1 (%): 82.550 -> 81.425 | No clear effect |
| combine rank and ssi losses instead of either alone (ablation) | PatchRefiner · CityScapes | Boundary Recall: 47.81 -> 48.78 | Improved |
| increase DSD loss weight lambda from 0.1 toward 1 (ablation) | PatchRefiner · CityScapes | Boundary Recall: 48.78 -> 58.48 | Mixed |
| use DSD loss versus standard fine-tuning | PatchRefiner · ScanNet++ | RMS: 0.268 -> 0.268 | No clear effect |
Closest papers by meaning
- PatchFusion: An End-to-End Tile-Based Framework for High-Resolution Monocular Metric Depth Estimation (CVPR 2024 · similarity 0.90)
- SharpDepth: Sharpening Metric Depth Predictions Using Diffusion Distillation (CVPR 2025 · similarity 0.86)
- DepthFM: Fast Generative Monocular Depth Estimation with Flow Matching (AAAI 2025 · similarity 0.86)
- Learning Deformable Hypothesis Sampling for Accurate PatchMatch Multi-View Stereo (AAAI 2024 · similarity 0.85)
- Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025 · similarity 0.85)
- BetterDepth: Plug-and-Play Diffusion Refiner for Zero-Shot Monocular Depth Estimation (NeurIPS 2024 · 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.
