Research map / Image generation / Super-resolution
Guided Depth Super-Resolution by Deep Anisotropic Diffusion
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 deep feature-guided anisotropic diffusion with source adjustment instead of LGR | guided depth super-resolution · Middlebury ×32 | MSE (cm²): 63.4 -> 47.6 | Helped |
| Use deep feature-guided anisotropic diffusion with source adjustment instead of LGR | guided depth super-resolution · NYUv2 ×32 | MSE (cm²): 253 -> 223 | Helped |
| Use deep feature-guided anisotropic diffusion with source adjustment instead of LGR | guided depth super-resolution · DIML ×32 | MSE (cm²): 57.0 -> 38.6 | Helped |
| Use deep feature-guided anisotropic diffusion with source adjustment instead of LGR | guided depth super-resolution · Middlebury ×8 | MSE (cm²): 7.26 -> 5.63 | Helped |
| Use deep feature-guided anisotropic diffusion with source adjustment instead of LGR | guided depth super-resolution · NYUv2 ×8 | MSE (cm²): 19.6 -> 17.1 | Helped |
| Use deep feature-guided anisotropic diffusion with source adjustment instead of LGR | guided depth super-resolution · DIML ×8 | MSE (cm²): 3.51 -> 2.87 | Helped |
| Replace ImageNet-pretrained ResNet-50 weights with random initialization (ablation) | DADA · Middlebury ×8 | MSE (cm²): 5.67 -> 5.92 | Hurt |
| Replace bicubic source initialization with constant initialization (ablation) | DADA · Middlebury ×8 | MSE (cm²): 5.67 -> 5.64 | No clear effect |
| Skip ×2 guide upsampling and feature-map downsampling (ablation) | DADA · Middlebury ×8 | MSE (cm²): 5.63 -> 5.92 | Hurt |
| Skip ×2 guide upsampling and feature-map downsampling (ablation) | DADA · NYUv2 ×8 | MSE (cm²): 16.6 -> 18.7 | Hurt |
| Skip ×2 guide upsampling and feature-map downsampling (ablation) | DADA · DIML ×8 | MSE (cm²): 2.93 -> 3.28 | Hurt |
| Increase the number of diffusion iterations (ablation) | DADA · Middlebury ×8 | MSE | Helped |
Closest papers by meaning
- SGNet: Structure Guided Network via Gradient-Frequency Awareness for Depth Map Super-resolution (AAAI 2024 · similarity 0.88)
- ResDiff: Combining CNN and Diffusion Model for Image Super-resolution (AAAI 2024 · similarity 0.87)
- Intrinsic Phase-Preserving Networks for Depth Super Resolution (AAAI 2024 · similarity 0.86)
- Effective Diffusion Transformer Architecture for Image Super-Resolution (AAAI 2025 · similarity 0.86)
- DORNet: A Degradation Oriented and Regularized Network for Blind Depth Super-Resolution (CVPR 2025 · similarity 0.86)
- Edge-SD-SR: Low Latency and Parameter Efficient On-device Super-Resolution with Stable Diffusion via Bidirectional Conditioning (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.