Research map / Image generation / Super-resolution
Joint Super-Resolution and Alignment of Tiny Faces
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 |
|---|---|---|---|
| Jointly train face alignment and super-resolution instead of using single-task baselines (ablation) | JASRNet · 300W | PSNR and NMSE: 23.41 -> 23.55; 5.71 -> 5.34 | Helped |
| Add intermediate feature fusion to the single-task baseline (ablation) | JASRNet · 300W | PSNR: 23.41 -> 23.50 | Helped |
| Add intermediate feature fusion to the single-task baseline (ablation) | JASRNet · 300W | NMSE: 5.71 -> 5.70 | No clear effect |
| Add intermediate feature fusion to joint training (ablation) | JASRNet · 300W | PSNR: 23.55 -> 23.58 | Helped |
| Add intermediate feature fusion to joint training (ablation) | JASRNet · 300W | NMSE: 5.34 -> 5.34 | No clear effect |
| Add the long skip connection to joint training with feature fusion (ablation) | JASRNet · 300W | PSNR and NMSE: 23.58 -> 23.69; 5.34 -> 5.26 | Helped |
| Use element-wise addition rather than concatenation for feature fusion (ablation) | JASRNet · 300W | PSNR and NMSE: 23.57 -> 23.69; 5.42 -> 5.29 | Helped |
| Use two alignment stages instead of one (ablation) | JASRNet · 300W | NMSE: 5.44 -> 5.36 | Helped |
| Increase feature extraction depth from 16 to 32 residual blocks (ablation) | JASRNet · 300W | PSNR and NMSE: 23.62 -> 23.69; 5.41 -> 5.29 | Helped |
| Use JASRNet rather than SuperFAN for tiny-face landmark localization | face alignment · 300W Full, 16x16 LR | NMSE: 6.55 -> 5.29 | Helped |
| Use JASRNet rather than SuperFAN for super-resolution | face SR · 300W | PSNR and SSIM: 23.13 -> 23.69; 0.691 -> 0.711 | Helped |
Closest papers by meaning
- Super-FAN: Integrated Facial Landmark Localization and Super-Resolution of Real-World Low Resolution Faces in Arbitrary Poses With GANs (CVPR 2018 · similarity 0.91)
- FSRNet: End-to-End Learning Face Super-Resolution With Facial Priors (CVPR 2018 · similarity 0.89)
- AnchorFace: An Anchor-based Facial Landmark Detector Across Large Poses (AAAI 2021 · similarity 0.88)
- LAFS: Landmark-based Facial Self-supervised Learning for Face Recognition (CVPR 2024 · similarity 0.86)
- SCE-MAE: Selective Correspondence Enhancement with Masked Autoencoder for Self-Supervised Landmark Estimation (CVPR 2024 · similarity 0.86)
- POPoS: Improving Efficient and Robust Facial Landmark Detection with Parallel Optimal Position Search (AAAI 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.