Research map / Image generation / Super-resolution
Adaptive Importance Learning for Improving Lightweight Image Super-Resolution Network
Lei Zhang, Peng Wang, Chunhua Shen, Lingqiao Liu, Wei Wei, Yanning Zhang, Anton van den Hengel
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 adaptive importance learning (AIL) | VDSR-f13 · Urban100 x2 | PSNR: 30.06 -> 30.42 | Improved |
| add adaptive importance learning | VDSR-f16 · Urban100 x2 | PSNR: 30.06 -> 30.53 | Improved |
| add adaptive importance learning | VDSR-f22 · Urban100 x2 | PSNR: 30.28 -> 30.64 | Improved |
| add adaptive importance learning | VDSR-f32 · Urban100 x2 | PSNR: 30.51 -> 30.84 | Improved |
| add adaptive importance learning | DRRN-f25 · Urban100 x2 | PSNR: 29.95 -> 30.58 | Improved |
| add adaptive importance learning | FSRCNN · Urban100 x2 | PSNR: 29.77 -> 30.24 | Improved |
| initialize importance from teacher instead of zeros (ablation) | VDSR-f13+AIL · Urban100 x3 | PSNR: 26.60 -> 26.76 | Improved |
| initialize importance from teacher instead of random values (ablation) | VDSR-f13+AIL · Urban100 x3 | PSNR: 26.66 -> 26.76 | Improved |
| add adaptive importance learning with zero initialization (no teacher) (ablation) | VDSR-f13 · Urban100 x3 | PSNR: 26.42 -> 26.60 | Improved |
| replace AIL with knowledge distillation from teacher output (ablation) | VDSR-f13 · Set5 x2 | PSNR: 37.18 -> 37.21 | No clear effect |
| replace AIL with knowledge distillation from teacher output (ablation) | VDSR-f13 · Set5 x4 | PSNR: 30.74 -> 30.68 | Hurt |
| add adaptive importance learning instead of distillation (ablation) | VDSR-f13 · Urban100 x3 | PSNR: 26.47 -> 26.76 | Improved |
Closest papers by meaning
- Uncertainty-Driven Loss for Single Image Super-Resolution (NeurIPS 2021 · similarity 0.89)
- Learning a Single Network for Scale-Arbitrary Super-Resolution (ICCV 2021 · similarity 0.89)
- Fast, Accurate, and Lightweight Super-Resolution with Cascading Residual Network (ECCV 2018 · similarity 0.89)
- Lightweight Image Super-Resolution via Flexible Meta Pruning (ICML 2024 · similarity 0.87)
- SAVSR: Arbitrary-Scale Video Super-Resolution via a Learned Scale-Adaptive Network (AAAI 2024 · similarity 0.87)
- Uncertainty-Aware GAN for Single Image Super Resolution (AAAI 2024 · similarity 0.87)
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.
