AI Paper Atlas

Research map / Image generation / GANs

DE-GAN: A Conditional Generative Adversarial Network for Document Enhancement

Mohamed Ali Souibgui, Yousri Kessentini

TPAMI 2020 · Image generation › GANs · approach: GAN latent optimization and evaluation · 174 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 adversarial training (DE-GAN) to the U-net generator (ablation)U-net document cleaner · Noisy OfficePSNR: 36.02 -> 38.12Improved
add adversarial training (DE-GAN) to the U-net generator (ablation)U-net document cleaner · Noisy OfficeSSIM: 0.9970 -> 0.9986Improved
use DE-GAN instead of the strongest prior method Vo et al.DE-GAN · DIBCO 2013F-measure: 94.4 -> 99.5Improved
use DE-GAN instead of the DIBCO 2017 winnerDE-GAN · DIBCO 2017F-measure: 91.04 -> 97.91Improved
use DE-GAN instead of the H-DIBCO 2018 winner, which uses pre- and post-processingDE-GAN · H-DIBCO 2018F-measure: 88.34 -> 77.59Hurt
use DE-GAN instead of the best prior watermark removal method (Dekel et al.)DE-GAN · Watermark removalPSNR: 36.02 -> 40.98Improved
use DE-GAN instead of cycleGANDE-GAN · H-DIBCO 2018F-measure: 56.33 -> 77.59Improved
use DE-GAN instead of pix2pix-HDDE-GAN · H-DIBCO 2018F-measure: 72.79 -> 77.59Improved
use DE-GAN instead of pix2pix-HD on small 300x300 patchesDE-GAN · Document deblurringPSNR: 19.89 -> 20.37Improved
use GAN models instead of a plain CNN for deblurringdeblurring · Document deblurringPSNR: 19.36 -> 20.37Improved
apply DE-GAN enhancement before Tesseract OCRTesseract OCR · 4 degraded/watermarked imagescharacter error rate: 0.37 -> 0.01Improved

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