AI Paper Atlas

Research map / Image generation / Image restoration

Self-Supervised Image Restoration with Blurry and Noisy Pairs

Zhilu Zhang, RongJian Xu, Ming Liu, Zifei Yan, Wangmeng Zuo

NeurIPS 2022 · Image generation › Image restoration · approach: Raw and low-light noise modeling · 7 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
jointly use the short-exposure noisy and long-exposure blurry images for restorationSelfIR · Synthetic sRGB Gaussian noisePSNR: 1.45 dBImproved
jointly use the short-exposure noisy and long-exposure blurry images for restorationSelfIR · Synthetic sRGB Poisson noisePSNR: 1.59 dBImproved
jointly use the short-exposure noisy and long-exposure blurry images for restorationSelfIR · Synthetic raw-RGB sensor noisePSNR: 1.69 dBImproved
replace clear-image supervision with noisy-image supervision for deblurring (ablation)Baseline_B · Synthetic deblurring with Gaussian, Poisson, and sensor noisePSNR: 28.24 -> 28.29; 28.24 -> 28.23; 28.14 -> 28.16No clear effect
add blurry-image auxiliary supervision to Neighbor2Neighbor (ablation)Neighbor2Neighbor · Synthetic sRGB Gaussian noisePSNR: 34.29 -> 34.45Improved
add auxiliary loss (ablation)SelfIR · Synthetic sRGB Gaussian noisePSNR: 35.65 -> 35.74Improved
increase regularization-loss weight from 0 to 2 (ablation)SelfIR · Synthetic sRGB Gaussian noisePSNR: 35.20 -> 35.74Improved
increase regularization-loss weight from 2 to 8 (ablation)SelfIR · Synthetic sRGB Gaussian noisePSNR: 35.74 -> 35.73No clear effect
increase auxiliary-loss weight from 0 to 2 (ablation)SelfIR · Synthetic sRGB Gaussian noisePSNR: 35.65 -> 35.74Improved
increase auxiliary-loss weight from 2 to 8 (ablation)SelfIR · Synthetic sRGB Gaussian noisePSNR: 35.74 -> 35.67Hurt
fine-tune on the real-world training pairsSelfIR · Real-world raw-RGB imagesPI: 0.55Improved

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