AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

United Defocus Blur Detection and Deblurring via Adversarial Promoting Learning

Wenda Zhao, Fei Wei, You He, Huchuan Lu

ECCV 2022 · Image generation › Autoregressive and latent generation · approach: Generalizable fake image detection · 20 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 three rounds of mutual adversarial promotion (Gws O3) to single detection (ablation)Gws · CUHK defocus detectionFmax: 0.785 -> 0.831Improved
add three rounds of mutual adversarial promotion (Gws O3) to single detection (ablation)Gws · CUHK defocus detectionMAE: 0.173 -> 0.125Improved
add three rounds of mutual adversarial promotion (Gws O3) to single detection (ablation)Gws · DUT defocus detectionFmax: 0.679 -> 0.722Improved
second optimization round of Gws after the first (ablation)Gws · DUT defocus detectionFmax: 0.696 -> 0.682Mixed
add three rounds of mutual promotion (Gsr O3) to single deblurring (ablation)Gsr · DP dataset deblurringPSNR: 22.02 -> 25.71Improved
add self-reference by directly combining focused and defocused images (DC-GAN) (ablation)U-GAN · DP dataset deblurringPSNR: 22.02 -> 24.16Improved
add UFAT instead of directly combining focused and defocused images (ablation)DC-GAN · DP dataset deblurringPSNR: 24.16 -> 25.71Improved
add UFAT instead of concatenating intermediate deep features (ablation)CDF-GAN · DP dataset deblurringPSNR: 24.80 -> 25.71Improved
add local-contrast blur prior L1 loss to adversarial supervision of Gws (ablation)Gws · defocus detectiondetection qualityImproved
replace fully-supervised KPAC with the weakly-supervised proposed modelKPAC · DP dataset deblurringPSNR: 26.51 -> 25.71Hurt
use the proposed method instead of SGNetSGNet · CUHK defocus detectionMAE: 0.199 -> 0.125Improved

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