Research map / Image generation / Super-resolution
Joint Face Hallucination and Deblurring via Structure Generation and Detail Enhancement
Yibing Song, Jiawei Zhang, Lijun Gong, Shengfeng He, Linchao Bao, Jinshan Pan, Qingxiong Yang, Ming–Hsuan Yang
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 local residual blocks to the baseline (ablation) | baseline B · Multi-PIE ablation | PSNR (SR / Deblur): 33.90 / 23.97 -> 33.95 / 24.02 | No clear effect |
| add facial mask input (ablation) | BL · Multi-PIE ablation | PSNR (SR / Deblur): 33.95 / 24.02 -> 34.22 / 24.83 | Improved |
| add dilation integration (ablation) | BL+M · Multi-PIE ablation | PSNR (SR / Deblur): 34.22 / 24.83 -> 34.63 / 25.31 | Improved |
| add detail enhancement with exemplar K-NN regression (ablation) | BLD+M · Multi-PIE ablation | PSNR (SR / Deblur): 34.63 / 25.31 -> 34.93 / 25.75 | Improved |
| add detail enhancement (ablation) | BLD+M · Multi-PIE ablation | SSIM (SR / Deblur): 0.91 / 0.85 -> 0.92 / 0.86 | Improved |
| use patch size 20x20 instead of 10x10 in K-NN search (ablation) | detail enhancement · Multi-PIE patch size | PSNR (SR / Deblur): 34.14 / 24.62 -> 34.93 / 25.75 | Improved |
| use patch size 30x30 instead of 20x20 (ablation) | detail enhancement · Multi-PIE patch size | PSNR (SR / Deblur): 34.93 / 25.75 -> 34.42 / 25.21 | Hurt |
| use 5 candidate patches instead of 3 (ablation) | detail enhancement · Multi-PIE patch number | PSNR (SR / Deblur): 34.05 / 24.97 -> 34.93 / 25.75 | Improved |
| use 7 candidate patches instead of 5 (ablation) | detail enhancement · Multi-PIE patch number | PSNR (SR / Deblur): 34.93 / 25.75 -> 34.92 / 25.75 | No clear effect |
| increase facial mask deviation from 0 to 10 pixels (ablation) | proposed method · Multi-PIE deblurring | PSNR: 25.54 -> 24.12 | Hurt |
| use proposed method instead of VDSR | VDSR · Multi-PIE deblurring | PSNR: 24.33 -> 25.75 | Improved |
| use proposed method instead of VDSR | VDSR · PubFig super-resolution | PSNR: 31.67 -> 31.86 | Improved |
Closest papers by meaning
- Exploiting Semantics for Face Image Deblurring (IJCV 2020 · similarity 0.91)
- Deep Semantic Face Deblurring (CVPR 2018 · similarity 0.91)
- Learning to Deblur Images with Exemplars (TPAMI 2018 · similarity 0.90)
- Learning to Deblur Face Images via Sketch Synthesis (AAAI 2020 · similarity 0.90)
- Semantic Face Hallucination: Super-Resolving Very Low-Resolution Face Images with Supplementary Attributes (TPAMI 2019 · similarity 0.89)
- Low-Light Face Super-resolution via Illumination, Structure, and Texture Associated Representation (AAAI 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.
