Research map / Image generation / Super-resolution
Spatial-Frequency Mutual Learning for Face Super-Resolution
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 |
|---|---|---|---|
| combine local spatial features with global frequency features using SFMNet | face super-resolution · CelebA ×4 and ×8; Helen ×4 and ×8 | PSNR: 31.88 -> 32.01; 27.42 -> 27.56; 32.41 -> 32.51; 27.08 -> 27.22 | Helped |
| remove the frequency branch (ablation) | SFMNet · CelebA ×8; Helen ×8 | PSNR: 27.35 -> 27.56; 26.98 -> 27.22 | Hurt |
| replace FSIB with concatenation followed by convolutional layers (ablation) | SFMNet · CelebA ×8; Helen ×8 | PSNR: 27.39 -> 27.56; 27.01 -> 27.22 | Hurt |
| replace SFCA with concatenation followed by a convolutional layer (ablation) | SFMNet · CelebA ×8; Helen ×8 | PSNR: 27.40 -> 27.56; 27.10 -> 27.22 | Hurt |
| add a spatial discriminator (ablation) | GAN-based SFMNet · CelebA ×8; Helen ×8 | LPIPS: 0.0869 -> 0.0684; 0.1061 -> 0.0847 | Helped |
| add a frequency discriminator to the spatial discriminator (ablation) | GAN-based SFMNet · CelebA ×8; Helen ×8 | LPIPS: 0.0684 -> 0.0594; 0.0847 -> 0.0760 | Helped |
| add a frequency discriminator to the spatial discriminator (ablation) | GAN-based SFMNet · CelebA ×8; Helen ×8 | NIQE: 6.931 -> 6.690; 7.475 -> 7.020 | Helped |
| train with the GAN-based model instead of the PSNR-oriented model | SFMNet · CelebA ×4 and ×8; Helen ×4 and ×8 | PSNR: 32.01 -> 30.99; 27.56 -> 26.48; 32.51 -> 31.54; 27.22 -> 26.39 | Hurt |
Closest papers by meaning
- Low-Light Face Super-resolution via Illumination, Structure, and Texture Associated Representation (AAAI 2024 · similarity 0.86)
- Beyond Spatial Domain: Cross-domain Promoted Fourier Convolution Helps Single Image Dehazing (AAAI 2025 · similarity 0.85)
- Learning Spatially Collaged Fourier Bases for Implicit Neural Representation (AAAI 2024 · similarity 0.84)
- Diffusion Prior Interpolation for Flexibility Real-World Face Super-Resolution (AAAI 2025 · similarity 0.84)
- Misalignment-Robust Frequency Distribution Loss for Image Transformation (CVPR 2024 · similarity 0.84)
- SFDM: Robust Decomposition of Geometry and Reflectance for Realistic Face Rendering from Sparse-view Images (CVPR 2025 · 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.