Research map / Image generation / Super-resolution
Efficient Real-World Image Super-Resolution Via Adaptive Directional Gradient Convolution
Long Peng, Zhanfeng Feng, Renjing Pei, Wenbo Li, Jiaming Guo, Xueyang Fu, Yang Wang, Yang Hua Cao, Zheng-Jun Zha
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 VConv with DGConv | VDSR · Real-SR avg | PSNR: 23.65 -> 24.14 | Improved |
| replace VConv with DGConv | EDSR · Real-SR avg | PSNR: 24.99 -> 25.18 | Improved |
| replace VConv with DGConv | IMDN · Real-SR avg | PSNR: 24.11 -> 24.52 | Improved |
| replace VConv with DGConv | SRResNet · Real-SR avg | LPIPS: 0.352 -> 0.335 | Improved |
| use five parallel VConvs instead of one (ablation) | SRResNet · Real-SR avg | PSNR: 24.996 -> 24.77 | Hurt |
| use ECB block instead of VConv (ablation) | SRResNet · Real-SR avg | PSNR: 24.996 -> 24.85 | Hurt |
| add IDG convolution (ablation) | DGConv · Real-SR avg | PSNR: 25.21 -> 25.41 | Improved |
| add CSG convolution (ablation) | DGConv · Real-SR avg | LPIPS: 0.323 -> 0.309 | Improved |
| add VConv branch (ablation) | DGConv · Real-SR avg | PSNR: 25.38 -> 25.41 | Hurt |
| increase feature channels 32 -> 64 (ablation) | DGPNet · Real-SR avg | PSNR: 25.37 -> 25.41 | Improved |
| use local mean instead of central value (ablation) | DGConv on EDSR · Real-SR avg | PSNR: 24.947 -> 25.186 | Improved |
| replace learnable convolution with fixed handcrafted kernels (ablation) | SRResNet · Real-SR | PSNR: 0.87 dB drop | Hurt |
Closest papers by meaning
- D2C-SR: A Divergence to Convergence Approach for Real-World Image Super-Resolution (ECCV 2022 · similarity 0.90)
- Component Divide-and-Conquer for Real-World Image Super-Resolution (ECCV 2020 · similarity 0.90)
- LinearSR: Unlocking Linear Attention for Stable and Efficient Image Super-Resolution (ICLR 2026 · similarity 0.89)
- SAVSR: Arbitrary-Scale Video Super-Resolution via a Learned Scale-Adaptive Network (AAAI 2024 · similarity 0.88)
- Towards Realistic Data Generation for Real-World Super-Resolution (ICLR 2025 · similarity 0.88)
- Noise-Started One-Step Image Super-Resolution via LR-Conditioned SplitMeanFlow and GAN Refinement (NeurIPS 2026 · 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.
