Research map / 3D vision / Depth and stereo
Single Image Depth Estimation Trained via Depth From Defocus Cues
Check 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 |
|---|---|---|---|
| train with depth-from-defocus cues and the proposed PSF renderer instead of compositional rendering (ablation) | Depth estimation on KITTI · KITTI F1 | RMSE: 12.395 -> 5.187 | Helped |
| train with depth-from-defocus cues and the proposed PSF renderer instead of compositional rendering (ablation) | Depth estimation on KITTI · KITTI F2 | RMSE: 11.177 -> 4.233 | Helped |
| add dense connections and self-attention to ASPP (ablation) | KITTI focus-stack reconstruction · F1 | RMSE: 5.412 -> 5.187 | Helped |
| add dense connections and self-attention to ASPP (ablation) | KITTI focus-stack reconstruction · F10 | RMSE: 4.194 -> 4.186 | Helped |
| add dense connections to ASPP (ablation) | KITTI focus-stack reconstruction · F1 | RMSE: 5.412 -> 5.285 | Helped |
| add self-attention to ASPP (ablation) | KITTI focus-stack reconstruction · F1 | RMSE: 5.412 -> 5.387 | Helped |
| increase rendered focused training images from F1 to F10 (ablation) | Depth-from-defocus estimator · KITTI → NYU | Correlation: 0.121 ± 0.006 -> 0.429 ± 0.009 | Helped |
| increase rendered focused training images from F1 to F10 (ablation) | Depth-from-defocus estimator · NYU → D3Net | Correlation: 0.530 ± 0.048 -> 0.434 ± 0.052 | Mixed |
| use a far focus distance that increases depth of field and weakens focus cues (ablation) | Depth-from-defocus estimator · Make3D | accuracy | Hurt |
| train using depth-from-defocus cues rather than direct depth supervision | Monocular depth estimation · NYU → KITTI | Correlation: 0.456 ± 0.006 -> 0.634 ± 0.005 | Helped |
| train using depth-from-defocus cues rather than direct depth supervision | Monocular depth estimation · KITTI → NYU | Correlation: 0.423 ± 0.010 -> 0.429 ± 0.009 | No clear effect |
Closest papers by meaning
- Focus on Defocus: Bridging the Synthetic to Real Domain Gap for Depth Estimation (CVPR 2020 · similarity 0.90)
- Fully Self-Supervised Depth Estimation From Defocus Clue (CVPR 2023 · similarity 0.90)
- Unsupervised Monocular Depth Estimation With Left-Right Consistency (CVPR 2017 · similarity 0.90)
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.88)
- Blurry-Edges: Photon-Limited Depth Estimation from Defocused Boundaries (CVPR 2025 · similarity 0.87)
- Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data (CVPR 2024 · similarity 0.87)
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.