Research map / Image generation / Super-resolution
Unsupervised Real-World Super-Resolution: A Domain Adaptation Perspective
Wei Wang, Haochen Zhang, Zehuan Yuan, Changhu Wang
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 feature alignment loss L_align on top of the regularization losses (ablation) | Domain-adaptation SR network · NTIRE 2020 Track 1 | LPIPS: 0.373 -> 0.296 | Improved |
| add feature alignment loss L_align on top of the regularization losses (ablation) | Domain-adaptation SR network · NTIRE 2020 Track 1 | SSIM: 0.663 -> 0.690 | Improved |
| add feature alignment loss L_align on top of the regularization losses (ablation) | Domain-adaptation SR network · NTIRE 2020 Track 1 | PSNR: 25.53 -> 25.14 | Hurt |
| add feature domain regularization losses (L_res, L_sty, L_idt, L_cyc) to the alignment-only model (ablation) | Domain-adaptation SR network · NTIRE 2020 Track 1 | LPIPS: 0.448 -> 0.296 | Improved |
| add feature alignment loss alone to the base model (ablation) | Base model (L_rec only) · NTIRE 2020 Track 1 | LPIPS: 0.436 -> 0.448 | Hurt |
| add feature alignment loss alone to the base model (ablation) | Base model (L_rec only) · NTIRE 2020 Track 1 | PSNR: 25.65 -> 25.20 | Hurt |
| use full proposed framework instead of base model (ablation) | Base model (L_rec only) · NTIRE 2020 Track 1 | LPIPS: 0.436 -> 0.296 | Improved |
| train on source LR with complex degradation instead of bicubic downsampling (ablation) | Base model · NTIRE 2020 Track 1 | LPIPS | Improved |
| add proposed domain adaptation framework to base model for either source LR synthesis (ablation) | Base model · NTIRE 2020 Track 1 | LPIPS | Improved |
Closest papers by meaning
- Unpaired Image Super-Resolution Using Pseudo-Supervision (CVPR 2020 · similarity 0.92)
- Unsupervised Real-World Image Super Resolution via Domain-Distance Aware Training (CVPR 2021 · similarity 0.92)
- Kernel Modeling Super-Resolution on Real Low-Resolution Images (ICCV 2019 · similarity 0.91)
- Dual Adversarial Adaptation for Cross-Device Real-World Image Super-Resolution (CVPR 2022 · similarity 0.91)
- Learning a Single Convolutional Super-Resolution Network for Multiple Degradations (CVPR 2018 · similarity 0.91)
- Low-Res Leads the Way: Improving Generalization for Super-Resolution by Self-Supervised Learning (CVPR 2024 · similarity 0.89)
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.
