AI Paper Atlas

Research map / Image generation / Super-resolution

Toward Real-World Super-Resolution via Adaptive Downsampling Models

Sanghyun Son, Jaeha Kim, Wei‐Sheng Lai, Ming–Hsuan Yang, Kyoung Mu Lee

TPAMI 2021 · Image generation › Super-resolution · approach: Real-world super-resolution · 48 citations (OpenAlex)

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 changedWhereMeasuredEffect
use LFL data term instead of L1 to bicubic-downsampled HR (ablation)unsupervised downsampler · x2 k4 LR PSNRPSNR between LR and Down: 37.22 -> 43.75Improved
use ADL instead of LFL data term (ablation)unsupervised downsampler · x4 k0 LR PSNRPSNR between LR and Down: 31.13 -> 38.24Improved
train SR on ADL-downsampled images instead of bicubic (ablation)EDSR · synthetic DIV2K x2 k4PSNR: 27.95 -> 32.05Improved
use ADL instead of LFL for the downsampler (ablation)EDSR · synthetic DIV2K x2 k4PSNR: 31.57 -> 32.05Improved
use a larger RRDB SR backbone with ADL (ablation)ADL · synthetic DIV2K x4 k4PSNR: 26.97 -> 27.19Improved
use ADL instead of LFL (ablation)EDSR · RealSR-V3 Nikon x4PSNR: 25.90 -> 26.46Improved
raise balancing weight alpha from 1 to 100 (ablation)ADL + EDSR · x2 k4PSNR: 29.08 -> 32.05Improved
raise balancing weight alpha from 1 to 200 (ablation)LFL + EDSR · x2 k4PSNR: 29.36 -> 31.57Improved
increase training LR images from 1 to 100 (ablation)LFL + EDSR · x2 k4PSNR: 25.43 -> 31.21Improved
add ADL over LFL with a single LR image (ablation)EDSR · x2 k4, 1 LR imagePSNR: 25.43 -> 29.84Improved
backpropagate SR loss into downsampler (end-to-end) (ablation)ADL + EDSR · x2 k4PSNR: 32.09 -> 31.39Hurt
add fake quantization to end-to-end training (ablation)ADL + EDSR · x2 k4PSNR: 31.39 -> 31.50No clear effect

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy