Research map / 3D vision / Depth and stereo
Depth from a Single Image by Harmonizing Overcomplete Local Network Predictions
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use all scales and orders of derivative coefficients (full overcomplete set) vs. restricted subsets (ablation) | global depth estimation · NYUv2 val subset (100 images) | RMSE (lin.): 0.6921 vs 0.7241-0.7971 for ablations | Helped |
| Add scale 3 derivatives (to scales 0,1,2) (ablation) | global depth estimation · NYUv2 val subset (100 images) | RMSE (lin.): 0.7241 -> 0.6921 | Helped |
| Add second order derivatives (to orders 0,1) (ablation) | global depth estimation · NYUv2 val subset (100 images) | RMSE (lin.): 0.6966 -> 0.6921 (small) | Helped |
| Use only zeroth order coefficients across all scales instead of full set (ablation) | global depth estimation · NYUv2 val subset (100 images) | RMSE (lin.): 0.7971 vs 0.6921 | Hurt |
| Combine multiple scales with only zeroth order derivatives vs pointwise depth distributions alone (ablation) | global depth estimation · NYUv2 val subset (100 images) | RMSE (lin.): 0.7971 (Order 0 all scales) vs 0.7424 (Scale 0 only) | Hurt |
| Overcomplete derivative-distribution representation with globalization (proposed) vs prior methods | monocular depth estimation · NYUv2 test set | RMSE (lin.): 0.620 vs 0.641 (Eigen 2015 VGG) | Helped |
Closest papers by meaning
- Generating and Exploiting Probabilistic Monocular Depth Estimates (CVPR 2020 · similarity 0.91)
- SDC-Depth: Semantic Divide-and-Conquer Network for Monocular Depth Estimation (CVPR 2020 · similarity 0.90)
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.88)
- DME: Unveiling the Bias for Better Generalized Monocular Depth Estimation (AAAI 2024 · similarity 0.88)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.88)
- Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025 · similarity 0.88)
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.