AI Paper Atlas

Research map / Image generation / Super-resolution

Learning to Super-Resolve Blurry Face and Text Images

Xiangyu Xu, Deqing Sun, Jinshan Pan, Yujin Zhang, Hanspeter Pfister, Ming-Hsuan Yang

ICCV 2017 · Image generation › Super-resolution · approach: CNN and transformer SR architectures · 237 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
add feature matching loss (G2D3) to basic GAN loss (ablation)SCGAN · Face datasetPSNR (dB): 22.82 -> 24.57Improved
use feature matching loss (G2D3) instead of pixel-wise loss (ablation)SCGAN · Face datasetPSNR (dB): 25.12 -> 24.57 (DCR 1.1331 -> 1.0687)Mixed
use discriminator features from layer 3 for D update (G2D3) instead of layer 2 (G2D2) (ablation)SCGAN · Face datasetPSNR (dB): 22.95 -> 24.57Improved
use layer 3 features for generator update (G3D3) instead of layer 2 (G2D3) (ablation)SCGAN · Face datasetPSNR (dB): 24.57 -> 23.17Hurt
split batches into real and generated when training discriminator (ablation)SCGAN · text and face imagesvisual qualityImproved
use bicubic interpolation instead of deconvolution for upsampling (ablation)SCGAN generator · synthetic face datasetPSNR (dB): 24.57 -> 23.45Hurt
train one multi-class model (MCGAN) instead of per-class SCGANGAN restoration · Text datasetPSNR (dB): 20.65 -> 20.12Hurt
train one multi-class model (MCGAN) instead of per-class SCGANGAN restoration · Face datasetPSNR (dB): 24.57 -> 23.95Hurt
use SCGAN instead of best sequential deblur+SR combinationblurry low-res restoration · Text datasetPSNR (dB): 16.45 -> 20.65Improved
use SCGAN instead of best sequential deblur+SR combinationblurry low-res restoration · Face datasetPSNR (dB): 22.17 -> 24.57Improved
use SCGAN instead of MSE-trained fine-tuned CNN ([18]+[15])blurry low-res restoration · Face datasetPSNR (dB): 22.60 -> 24.57Improved
use SCGAN instead of bicubic in subjective studyblurry low-res restoration · Text images user studymean opinion score: 1.10 -> 3.68Improved

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