Research map / 3D vision / Depth and stereo
Deep Ordinal Regression Network for Monocular Depth Estimation
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 ordinal regression with SID discretization instead of continuous-depth MSE regression (ablation) | DORN on KITTI · KITTI | δ < 1.25: 0.864 -> 0.915 | Helped |
| Discretize with SID instead of uniform discretization (ablation) | DORN on KITTI · KITTI | δ < 1.25: 0.900 -> 0.915 | Helped |
| Use ordinal regression rather than multiclass classification on SID-discretized depths (ablation) | DORN on KITTI · KITTI | δ < 1.25: 0.906 -> 0.915 | Helped |
| Apply SID discretization to MSE regression (ablation) | Depth regression on KITTI · KITTI | δ < 1.25: 0.864 -> 0.865 | Helped |
| Replace ordinal regression loss with BerHu loss (ablation) | ResNet DORN on KITTI · KITTI | δ < 1.25: 0.909 -> 0.932 | Hurt |
| Add a full-image encoder (ablation) | DORN on KITTI · KITTI | δ < 1.25: 0.906 -> 0.915 | Helped |
| Replace the proposed full-image encoder with an fc-fashion encoder (ablation) | DORN on KITTI · KITTI | δ < 1.25: 0.910 -> 0.915 | Hurt |
| Use the proposed full-image encoder rather than an fc-fashion encoder (ablation) | DORN on KITTI · KITTI | Params: 753$M$ -> 51$M$ | Helped |
| Use DORN instead of the official baseline | Monocular depth estimation · ScanNet online test | abs rel.: 0.25 -> 0.14 | Helped |
| Use DORN instead of the official baseline | Monocular depth estimation · KITTI online test | SILog: 18.19 -> 11.77 | Helped |
Closest papers by meaning
- Depth from a Single Image by Harmonizing Overcomplete Local Network Predictions (NeurIPS 2016 · similarity 0.88)
- SDC-Depth: Semantic Divide-and-Conquer Network for Monocular Depth Estimation (CVPR 2020 · similarity 0.88)
- Efficient Edge-Preserving Multi-View Stereo Network for Depth Estimation (AAAI 2023 · similarity 0.88)
- VA-DepthNet: A Variational Approach to Single Image Depth Prediction (ICLR 2023 · similarity 0.87)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.86)
- Scalable Autoregressive Monocular Depth Estimation (CVPR 2025 · 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.