AI Paper Atlas

Research map / Image generation / Super-resolution

Reference-Based Image Super-Resolution with Deformable Attention Transformer

Jiezhang Cao, Jingyun Liang, Kai Zhang, Yawei Li, Yulun Zhang, Wenguan Wang, Luc Van Gool

ECCV 2022 · Image generation › Super-resolution · approach: Efficient lightweight super-resolution · 113 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 reference-based deformable attention (RDA) instead of feature warping on the most relevant correspondence (ablation)DATSR · CUFED5PSNR/SSIMImproved
add residual feature aggregation (RFA) instead of plain CNNs (ablation)DATSR · CUFED5PSNR/SSIMImproved
replace RFA with CNNs but keep RDA vs C2-Matching (ablation)DATSR with CNNs · CUFED5PSNR/SSIMImproved
train DATSR with adversarial loss instead of reconstruction loss onlyDATSR · CUFED5PSNR: 28.72 -> 27.95Hurt
DATSR vs C2-Matching, reconstruction lossreference-based SR · CUFED5PSNR: 28.24 -> 28.72Improved
DATSR vs C2-Matching on L4 least similar referencesreference-based SR · CUFED5 L4PSNR: 26.85 -> 26.96Improved
DATSR vs C2-Matching on L1 most similar referencesreference-based SR · CUFED5 L1PSNR: 28.11 -> 28.50Improved
DATSR vs C2-Matching, all lossesreference-based SR · Manga109 vs WR-SRPSNR: 27.80 -> 27.87Improved
DATSR (random reference) vs SwinIR on Urban100reference-based SR · Urban100PSNR: 26.26 -> 26.52Improved
single reference DATSR vs multi-reference CIMR-SR and AMRSRDATSR · CUFED5PSNRImproved

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