AI Paper Atlas

Research map / Image generation / GANs

Global and Local Enhancement Networks for Paired and Unpaired Image Enhancement

Han-Ul Kim, Young Jun Koh, Chang-Su Kim

ECCV 2020 · Image generation › GANs · approach: GAN-based image manipulation · 77 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 local enhancement network (LEN) to GEN, paired (ablation)GEN · MIT-Adobe 5K (paired learning)PSNR / SSIM: PSNR 25.47 -> 25.88; SSIM 0.917 -> 0.925Improved
add LEN to GEN, unpaired (ablation)GEN · MIT-Adobe 5K (unpaired learning)PSNR / SSIM: PSNR 23.74 -> 23.82; SSIM 0.885 -> 0.889Improved
use channel-wise intensity transform instead of pixel-wise color predictionGEN · MIT-Adobe 5K (paired learning)PSNR / SSIM training speedImproved
add degradation GAN pseudo pairs to train LEN (vs adversarial loss only) (ablation)GEN & LEN · MIT-Adobe 5K (unpaired, training scheme ablation)PSNR: 21.78 -> 23.82Improved
add degradation GAN pseudo pairs to train LEN (vs adversarial loss only) (ablation)GEN & LEN · MIT-Adobe 5K (unpaired, training scheme ablation)SSIM: 0.847 -> 0.889Improved
replace cyclic consistency loss with proposed cyclic color loss, GEN (ablation)GEN · MIT-Adobe 5K (unpaired, training scheme ablation)PSNR: 23.11 -> 23.74Improved
replace cyclic consistency loss with proposed cyclic color loss, GEN & LEN (ablation)GEN & LEN · MIT-Adobe 5K (unpaired, training scheme ablation)PSNR: 23.18 -> 23.82Improved
add proposed two-stage training with degradation GAN to GEN (ablation)GEN · MIT-Adobe 5K (unpaired, training scheme ablation)PSNR: 23.05 -> 23.74Improved

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