Research map / Image generation / Super-resolution
Learning Scene Structure Guidance via Cross-Task Knowledge Transfer for Single Depth Super-Resolution
Baoli Sun, Xinchen Ye, Baopu Li, Haojie Li, Zhihui Wang, Rui Xu
Check 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 |
|---|---|---|---|
| Transfer cross-task knowledge from the auxiliary RGB depth-estimation task during training (ablation) | DSRNet · Middlebury ×8 | MAD: 0.398 -> 0.356 | Improved |
| Transfer cross-task knowledge from the auxiliary RGB depth-estimation task during training (ablation) | DSRNet · NYUv2 ×8 | RMSE: 3.13 -> 2.75 | Improved |
| Add output-space distillation (ablation) | DSRNet with DENet · Middlebury ×8 | MAD: 0.398 -> 0.377 | Improved |
| Add output-space distillation (ablation) | DSRNet with DENet · NYUv2 ×8 | RMSE: 3.13 -> 2.88 | Improved |
| Add affinity-space distillation (ablation) | DSRNet with DENet · Middlebury ×8 | MAD: 0.398 -> 0.384 | Improved |
| Add affinity-space distillation (ablation) | DSRNet with DENet · NYUv2 ×8 | RMSE: 3.13 -> 2.93 | Improved |
| Add both output-space and affinity-space distillation (ablation) | DSRNet with DENet · Middlebury ×8 | MAD: 0.398 -> 0.372 | Improved |
| Add both output-space and affinity-space distillation (ablation) | DSRNet with DENet · NYUv2 ×8 | RMSE: 3.13 -> 2.85 | Improved |
| Add structure prediction regularization (ablation) | DSRNet with both distillation losses · Middlebury ×8 | MAD: 0.372 -> 0.356 | Improved |
| Add structure prediction regularization (ablation) | DSRNet with both distillation losses · NYUv2 ×8 | RMSE: 2.85 -> 2.75 | Improved |
| Use uncertainty-induced attention instead of plain feature concatenation (ablation) | SPNet · Middlebury Art ×8 structure prediction | MAD: 1.403 -> 1.386 | Improved |
| Train the proposed single-depth method rather than DGDIE | DSRNet · NYUv2 ×4 | RMSE: 1.56 -> 1.49 | Improved |
Closest papers by meaning
- Structure Flow-Guided Network for Real Depth Super-resolution (AAAI 2023 · similarity 0.90)
- SGNet: Structure Guided Network via Gradient-Frequency Awareness for Depth Map Super-resolution (AAAI 2024 · similarity 0.89)
- Dual Graph Regularized Deep Unfolding Network for Guided Depth Map Super-resolution (CVPR 2026 · similarity 0.88)
- Scene Prior Filtering for Depth Super-Resolution (IJCV 2026 · similarity 0.88)
- Intrinsic Phase-Preserving Networks for Depth Super Resolution (AAAI 2024 · similarity 0.86)
- DORNet: A Degradation Oriented and Regularized Network for Blind Depth Super-Resolution (CVPR 2025 · similarity 0.85)
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.
