Research map / Image generation / Super-resolution
Towards Lightweight Super-Resolution With Dual Regression Learning
Yong Guo, Mingkui Tan, Zeshuai Deng, Jingdong Wang, Qi Chen, Jiezhang Cao, Yanwu Xu, Jian Chen
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 dual regression learning (ablation) | DRN-S · Set5 (4x SR) | PSNR: 32.53 -> 32.68 | Improved |
| add dual regression learning (ablation) | DRN-L · Urban100 (4x SR) | PSNR: 26.77 -> 27.03 | Improved |
| add dual regression to DAT training | DAT · Urban100 (4x SR) | PSNR: 27.87 -> 28.04 | Improved |
| add dual regression to SwinIR training | SwinIR · Urban100 (4x SR) | PSNR: 27.45 -> 27.80 | Improved |
| add dual regression to channel pruning (ablation) | DRN-S 30% pruning · Set5 (4x SR) | PSNR: 32.53 -> 32.66 | Improved |
| add dual regression to channel pruning (ablation) | DRN-S 50% pruning · Urban100 (4x SR) | PSNR: 26.47 -> 26.59 | Improved |
| add dual regression to channel pruning (ablation) | DRN-S 70% pruning · BSDS100 (4x SR) | PSNR: 27.63 -> 27.66 | Improved |
| add channel-wise weight sharing to DARTS search (ablation) | DARTS channel search · Set5 (4x SR) | search cost and PSNR: 23 -> 10 hours; 32.37 -> 32.40 | Improved |
| add importance-aware search (ablation) | DRC channel search · Set5 (4x SR) | PSNR: 32.40 -> 32.53 | Improved |
| add dual regression to channel number search (ablation) | DRC channel search · Set5 (4x SR) | PSNR: 32.53 -> 32.66 | Improved |
| replace single dual model with progressive dual models (ablation) | DRN-S · Set5 (4x SR) | PSNR: 32.63 -> 32.68 | Improved |
| use 2x larger dual models (ablation) | DRN-S · Set5 (4x SR) | PSNR: 32.68 -> 32.69 | No clear effect |
Closest papers by meaning
- Closed-Loop Matters: Dual Regression Networks for Single Image Super-Resolution (CVPR 2020 · similarity 0.95)
- Image Super-Resolution via Dual-State Recurrent Networks (CVPR 2018 · similarity 0.90)
- Accelerating the Super-Resolution Convolutional Neural Network (ECCV 2016 · similarity 0.89)
- Image Super-Resolution Using Deep Convolutional Networks (TPAMI 2016 · similarity 0.89)
- Lightweight Image Super-Resolution via Flexible Meta Pruning (ICML 2024 · similarity 0.89)
- Low-Res Leads the Way: Improving Generalization for Super-Resolution by Self-Supervised Learning (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.
