AI Paper Atlas

Research map / Image generation / Super-resolution

Memory-Augmented Deep Unfolding Network for Guided Image Super-resolution

Man Zhou, Keyu Yan, Jinshan Pan, Wenqi Ren, Qi Xie, Xiangyong Cao

IJCV 2022 · Image generation › Super-resolution · approach: CNN and transformer SR architectures · 92 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
increase unfolding stages from 1 to 4 (ablation)MADUNet · GaoFen2 stage studyPSNR: 41.2772 -> 41.8577Improved
increase unfolding stages from 4 to 5 (ablation)MADUNet · GaoFen2 stage studyPSNR: 41.8577 -> 41.7545Hurt
increase unfolding stages from 3 to 4 (ablation)MADUNet · GaoFen2 stage studySSIM: 0.9697 -> 0.9697No clear effect
use memory-augmented unfolding with non-local module versus SRPPNNMADUNet · WorldView-IIPSNR: 41.4538 -> 41.8577Improved
use model-driven unfolding versus GPPNNMADUNet · WorldView-IIPSNR: 41.1622 -> 41.8577Improved
compare against second-best SRPPNNMADUNet · GaoFen2PSNR: 47.1998 -> 47.2668Improved
apply pre-trained GaoFen2 model to full-resolution dataMADUNet · GaoFen2 full-resolutionD_s: 0.1139 vs 0.1087 (SFIM)No clear effect
compare against DKN on bicubic depth SRMADUNet · Depth SR bicubicaverage RMSE: 0.09 (4x), 0.36 (8x), 0.85 (16x) lowerImproved
compare against DKN on direct depth SRMADUNet · Depth SR directaverage RMSE: 0.15 (4x), 0.23 (8x), 0.19 (16x) lowerImproved
generalize NYU-trained model to Lu dataset at 16xMADUNet · Lu direct down-samplingRMSE: 0.03 lower than DKNNo clear effect
compare against MINet on MR x4MADUNet · MR image SR x4PSNR: 35.1998 -> 35.2928Improved
compare against MINet on MR x2MADUNet · MR image SR x2PSNR: 38.2553 -> 38.3110Improved

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