Research map / Image generation / Editing and style transfer
Progressive Semantic-Aware Style Transformation for Blind Face Restoration
Chaofeng Chen, Xiaoming Li, Lingbo Yang, Xianhui Lin, Lei Zhang, Kwan-Yee K. Wong
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 |
|---|---|---|---|
| Use PSFR-GAN instead of WaveletSRNet | Blind face restoration · CelebAHQ-Test | PSNR: 23.50 -> 23.64 | Improved |
| Use PSFR-GAN instead of WaveletSRNet | Blind face restoration · CelebAHQ-Test | SSIM: 0.6595 -> 0.6557 | Hurt |
| Use PSFR-GAN instead of DFDNet | Blind face restoration · CelebAHQ-Test | MSSIM: 0.7650 -> 0.7740 | Improved |
| Use PSFR-GAN instead of DFDNet | Blind face restoration · CelebAHQ-Test | LPIPS: 0.3791 -> 0.3042 | Improved |
| Use PSFR-GAN instead of DFDNet | Blind face restoration · CelebAHQ-Test | FID: 37.34 -> 23.20 | Improved |
| Use PSFR-GAN instead of DFDNet | Blind face restoration · PSFR-RealTest | FID: 37.63 -> 30.39 | Improved |
| Add parsing-map guidance (ablation) | PSFR-GAN with LQ input only · PSFR-RealTest | FID: 47.48 -> 32.37 | Improved |
| Add semantic-aware style loss (ablation) | PSFR-GAN with LQ input only · PSFR-RealTest | FID: 47.48 -> 34.22 | Improved |
| Add both parsing-map guidance and semantic-aware style loss (ablation) | PSFR-GAN with LQ input only · PSFR-RealTest | FID: 47.48 -> 30.39 | Improved |
| Remove parsing-map guidance (ablation) | Full PSFR-GAN · PSFR-RealTest | FID: 30.39 -> 34.22 | Hurt |
| Remove semantic-aware style loss (ablation) | Full PSFR-GAN · PSFR-RealTest | FID: 30.39 -> 32.37 | Hurt |
| Add high-resolution face pixel supervision to parsing-network training (ablation) | LQ face parsing network · Natural low-resolution face parsing | parsing quality | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- WaveletSRNet — Wavelet-SRNet: A Wavelet-Based CNN for Multi-Scale Face Super Resolution (ICCV 2017) · better on 1 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Towards Real-World Blind Face Restoration With Generative Facial Prior (CVPR 2021 · similarity 0.90)
- A Natural Language Guided Approach for Blind Face Restoration: Methodology and Dataset (TPAMI 2026 · similarity 0.90)
- Overcoming False Illusions in Real-World Face Restoration with Multi-Modal Guided Diffusion Model (ICLR 2025 · similarity 0.87)
- Learning Degradation-unaware Representation with Prior-based Latent Transformations for Blind Face Restoration (CVPR 2024 · similarity 0.87)
- ReF-LDM: A Latent Diffusion Model for Reference-based Face Image Restoration (NeurIPS 2024 · similarity 0.87)
- $InterLCM$: Low-Quality Images as Intermediate States of Latent Consistency Models for Effective Blind Face Restoration (ICLR 2025 · similarity 0.86)
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.
