Research map / 3D vision / Depth and stereo
Reconstruction-based Pairwise Depth Dataset for Depth Image Enhancement Using CNN
Junho Jeon, Seungyong Lee
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 |
|---|---|---|---|
| filter misaligned training pairs by SSIM (< 0.8) and large holes (ablation) | LapDEN · ScanNet training | training stability / loss | Improved |
| add dataset filtering of misaligned pairs (ablation) | LapDEN · ScanNet patches (qualitative) | edge sharpness | Improved |
| add gradient-based structure preserving loss L_S (ablation) | LapDEN · ScanNet patches (qualitative) | edge sharpness | Improved |
| use LapDEN instead of rolling guidance filter | RGF · ScanNet test | SSIM: 0.9065 -> 0.9229 | Improved |
| use LapDEN instead of rolling guidance filter | RGF · ScanNet test | RMSE: 0.2401 -> 0.2148 | Improved |
| use LapDEN instead of mutual-structure joint filter | Joint filter · ScanNet test | SSIM: 0.9152 -> 0.9229 | Improved |
| use LapDEN instead of mutual-structure joint filter | Joint filter · ScanNet test | RMSE: 0.2360 -> 0.2148 | Improved |
| enhance raw depth with LapDEN | raw depth · ScanNet test | RMSE: 0.3450 -> 0.2148 | Improved |
| pre-filter depth stream with LapDEN before dense 3D reconstruction | BundleFusion · ScanNet scenes (qualitative) | convergence speed | Improved |
| pre-filter depth stream and skip every other frame | BundleFusion · ScanNet scenes (qualitative) | reconstruction quality | No clear effect |
Closest papers by meaning
- Learning Joint 2D-3D Representations for Depth Completion (ICCV 2019 · similarity 0.86)
- StereoDRNet: Dilated Residual StereoNet (CVPR 2019 · similarity 0.85)
- Deep Depth Completion of a Single RGB-D Image (CVPR 2018 · similarity 0.85)
- Depth Coefficients for Depth Completion (CVPR 2019 · similarity 0.85)
- Self-Supervised Deep Depth Denoising (ICCV 2019 · similarity 0.85)
- A Two-Streamed Network for Estimating Fine-Scaled Depth Maps From Single RGB Images (ICCV 2017 · similarity 0.84)
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.
