Research map / Image generation / Super-resolution
Learning Frequency-Aware Dynamic Network for Efficient Super-Resolution
Wenbin Xie, Dehua Song, Chang Xu, Chunjing Xu, Hui Zhang, Yunhe 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 |
|---|---|---|---|
| replace EDSR with frequency-aware dynamic network (FAD-EDSR) | EDSR · Set5 x2 | FLOPs (G): 1338.8 -> 408.4 | Improved |
| replace EDSR with FAD-EDSR | EDSR · Set5 x2 | PSNR (dB): 38.11 -> 38.21 | Improved |
| replace RCAN with FAD-RCAN | RCAN · Urban100 x4 | FLOPs (G): 1452.5 -> 891.2 | Improved |
| use FAD-EDSR instead of AdaEDSR | AdaEDSR · Urban100 x4 | PSNR (dB): 26.58 -> 26.70 | Improved |
| deploy FAD-EDSR on single core CPU | EDSR · CPU latency (single core) | latency (s): 39.60s -> 25.44s | Improved |
| deploy FAD-EDSR on double core CPU | EDSR · CPU latency (double core) | latency (s): 20.08s -> 13.74s | Improved |
| use DCT frequency mask instead of random masks per image (ablation) | FADN (EDSR-baseline) · Urban100 x2 | PSNR (dB): 31.17 -> 31.72 | Improved |
| use random masks per block instead of per image (ablation) | FADN (EDSR-baseline) · Urban100 x2 | PSNR (dB): 31.17 -> 31.47 | Improved |
| use learnable mask instead of DCT mask (no guidance) (ablation) | FADN (EDSR-baseline) · Urban100 x2 | PSNR (dB): 31.72 -> 31.81 | Improved |
| add DCT guidance to learnable mask (ablation) | FADN (EDSR-baseline) · Urban100 x2 | PSNR (dB): 31.81 -> 31.90 | Improved |
| increase branches from 2 to 3 (ablation) | FADN (EDSR-baseline) · Urban100 x2 | PSNR (dB): 31.72 -> 31.85 | Improved |
| increase branches from 3 to 4 (ablation) | FADN (EDSR-baseline) · Urban100 x2 | PSNR (dB): 31.85 -> 31.88 | No clear effect |
Closest papers by meaning
- Dynamic High-Pass Filtering and Multi-Spectral Attention for Image Super-Resolution (ICCV 2021 · similarity 0.90)
- Learning Omni-Frequency Region-adaptive Representations for Real Image Super-Resolution (AAAI 2021 · similarity 0.90)
- FSR: A General Frequency-Oriented Framework to Accelerate Image Super-resolution Networks (AAAI 2023 · similarity 0.89)
- SAVSR: Arbitrary-Scale Video Super-Resolution via a Learned Scale-Adaptive Network (AAAI 2024 · similarity 0.86)
- ResDiff: Combining CNN and Diffusion Model for Image Super-resolution (AAAI 2024 · similarity 0.86)
- Exploring the Low-Pass Filtering Behavior in Image Super-Resolution (ICML 2024 · similarity 0.86)
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.
