Research map / 3D vision / Depth and stereo
Improving Monocular Depth Estimation by Leveraging Structural Awareness and Complementary Datasets
Tian Chen, Shijie An, Yuan Zhang, Chongyang Ma, Huayan Wang, Xiaoyan Guo, Wen Zheng
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 edge-aware consistency (ablation) | baseline Mb · NYUDv2 | RMSE: 0.430 -> 0.426 | Improved |
| add spatial attention blocks (SAB) (ablation) | Me · NYUDv2 | RMSE: 0.426 -> 0.417 | Improved |
| add global focal relative loss (GFRL) (ablation) | Me,SAB · NYUDv2 | RMSE: 0.417 -> 0.407 | Improved |
| use GFRL instead of original relative loss (ablation) | Me,SAB · NYUDv2 | delta1 accuracy: 0.864 -> 0.876 | Improved |
| use SAB instead of sSE (ablation) | Me,Lr · NYUDv2 | RMSE: 0.433 -> 0.407 | Improved |
| use SAB instead of CBAM (ablation) | Me,Lr · NYUDv2 | RMSE: 0.432 -> 0.407 | Improved |
| add HC Depth dataset to training (ablation) | Mfull · NYUDv2 | delta1 accuracy: 0.876 -> 0.885 | Improved |
| train on all six datasets (ablation) | Mfull · NYUDv2 | RMSE: 0.407 -> 0.376 | Improved |
| train on all six datasets (ablation) | Mfull · TUM (zero-shot) | RMSE: 1.109 -> 0.912 | Improved |
| add HC Depth dataset only (ablation) | Mfull · HC Depth test | delta1 accuracy: 0.600 -> 0.825 | Improved |
| train on all six datasets (ablation) | Mfull · HC Depth test | delta1 accuracy: 0.600 -> 0.879 | Improved |
| add synthetic and indoor datasets 2,3 only (ablation) | Mfull · HC Depth test | delta1 accuracy: 0.600 -> 0.610 | Improved |
Closest papers by meaning
- Guiding Monocular Depth Estimation Using Depth-Attention Volume (ECCV 2020 · similarity 0.91)
- Depth from a Single Image by Harmonizing Overcomplete Local Network Predictions (NeurIPS 2016 · similarity 0.90)
- NDDepth: Normal-Distance Assisted Monocular Depth Estimation (ICCV 2023 · similarity 0.90)
- Single View Stereo Matching (CVPR 2018 · similarity 0.90)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.90)
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.89)
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.
