AI Paper Atlas

Research map / Image generation / Image restoration

Flare7K++: Mixing Synthetic and Real Datasets for Nighttime Flare Removal and Beyond

Yuekun Dai, Chongyi Li, Shangchen Zhou, Ruicheng Feng, Yihang Luo, Chen Change Loy

TPAMI 2024 · Image generation › Image restoration · approach: Reflection removal · 68 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 Flare-R real flare data to Flare7K training (ablation)Uformer · Flare7K++ real-captured test setPSNR: 27.257 -> 27.633Improved
add Flare-R real flare data to Flare7K training (ablation)Uformer · Flare7K++ real-captured test setS-PSNR: 21.294 -> 22.603Improved
add Flare-R real flare data to Flare7K training (ablation)Uformer · Flare7K++ real-captured test setLPIPS: 0.0471 -> 0.0428Improved
add light source annotations to training (ablation)Uformer · Flare7K++ real-captured test setPSNR: 26.850 -> 27.633Improved
add light source annotations to training (ablation)Uformer · Flare7K++ real-captured test setS-PSNR: 21.909 -> 22.603Improved
add light source annotations to training (ablation)Uformer · Flare7K++ real-captured test setSSIM: 0.895 -> 0.894No clear effect
use end-to-end pipeline without subtraction instead of Wu et al. pipeline (ablation)Uformer · Flare7K++ real-captured test setPSNR: 27.166 -> 27.633Improved
use end-to-end pipeline without subtraction instead of Wu et al. pipeline (ablation)Uformer · Flare7K++ real-captured test setSSIM: 0.861 -> 0.894Improved
use end-to-end pipeline without subtraction instead of Wu et al. pipeline (ablation)Uformer · Flare7K++ real-captured test setS-PSNR: 22.118 -> 22.603Improved
train U-Net on Flare7K++ instead of Wu et al. datasetU-Net · Flare7K++ real-captured test setPSNR: 24.613 -> 27.189Improved
train on Flare7K++ with new pipeline instead of Flare7K baselineUformer · Flare7K++ real-captured test setS-PSNR: 21.563 -> 22.603Improved
use Uformer backbone instead of U-Net on Flare7K++Flare7K++ network · Flare7K++ real-captured test setPSNR: 27.189 -> 27.633Improved

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