Research map / Image generation / Super-resolution
Pairwise Distance Distillation for Unsupervised Real-World Image Super-Resolution
Yuehan Zhang, Seungjun Lee, Angela Yao
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 |
|---|---|---|---|
| apply PDD to RealESRGAN | RealESRGAN · RealSR-Canon | PSNR: 24.74 -> 25.94 | Improved |
| apply PDD to RealESRGAN | RealESRGAN · DRealSR-Panasonic | NRQM: 5.3751 -> 5.3540 | Hurt |
| apply PDD to RealESRGAN | RealESRGAN · RealSR-Canon | NRQM: 6.0649 -> 6.0647 | No clear effect |
| apply PDD to BSRGAN | BSRGAN · RealSR-Canon | SSIM: -0.006 | Hurt |
| naive distillation from generalist instead of PDD (ablation) | specialist model · NTIRE20 / RealSR | LPIPS, NRQM | Hurt |
| use EMA-updated generalist rather than fixed generalist (ablation) | PDD with RealESRGAN · RealSR-Canon | PSNR: 25.71 -> 25.94 | Improved |
| use EMA-updated generalist rather than fixed generalist (ablation) | PDD with RealESRGAN · NTIRE20-Valid | NRQM: 6.0938 -> 6.2263 | Improved |
| use cross-entropy instead of l1 for R_intra (ablation) | PDD with RealESRGAN · DRealSR-Olympus | NRQM: 4.9602 -> 5.3722 | Improved |
| use cross-entropy instead of l1 for R_intra (ablation) | PDD with RealESRGAN · RealSR-Canon | PSNR: 26.17 -> 25.94 | Hurt |
| combine R_intra and R_inter and add GAN loss (ablation) | PDD · RealSR / NTIRE20 | PSNR, NRQM | Mixed |
| apply PDD to specialist | bicubic specialist ESRGAN · DIV2K vs NTIRE20 predictions | KL divergence of low-level features | Improved |
Closest papers by meaning
- Toward Real-World Single Image Super-Resolution: A New Benchmark and a New Model (ICCV 2019 · similarity 0.90)
- Learning Many-to-Many Mapping for Unpaired Real-World Image Super-Resolution and Downscaling (TPAMI 2024 · similarity 0.90)
- Unsupervised Diffusion-Based Degradation Modeling for Real-World Super-Resolution (AAAI 2025 · similarity 0.89)
- Low-Res Leads the Way: Improving Generalization for Super-Resolution by Self-Supervised Learning (CVPR 2024 · similarity 0.89)
- Towards Realistic Data Generation for Real-World Super-Resolution (ICLR 2025 · similarity 0.89)
- Learning Coupled Dictionaries from Unpaired Data for Image Super-Resolution (CVPR 2024 · similarity 0.88)
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.
