AI Paper Atlas

Research map / Image generation / Image restoration

Adaptive Unfolding Total Variation Network for Low-Light Image Enhancement

Chuanjun Zheng, Daming Shi, Wentian Shi

ICCV 2021 · Image generation › Image restoration · approach: Raw and low-light noise modeling · 130 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 the learned noise level map M (ablation)UTVNet · sRGB-SID testPSNR (dB): 21.858 -> 22.691Improved
learn per-pixel noise level map M rather than a global noise variation only (ablation)UTVNet · sRGB-SID testPSNR (dB): 21.483 -> 22.691Improved
add global noise variation epsilon as base in the NLI-Block (ablation)UTVNet · sRGB-SID testPSNR (dB): 21.909 -> 22.691Improved
use modified activation A() (negatives set to epsilon) instead of ReLU (ablation)UTVNet · sRGB-SID testPSNR (dB): 22.136 -> 22.691Improved
increase unfolding iterations from 4 to 8 (ablation)UTVNet · sRGB-SID testPSNR (dB): 22.476 -> 22.691Improved
increase unfolding iterations from 8 to 10 (ablation)UTVNet · sRGB-SID testPSNR (dB): 22.691 -> 22.647Hurt
increase unfolding iterations from 8 to 12 (ablation)UTVNet · sRGB-SID testPSNR (dB): 22.691 -> 22.521Hurt
increase iterations from 8 to 12 on LPIPS (ablation)UTVNet · sRGB-SID testLPIPS: 0.3417 -> 0.3368Improved
use UTVNet instead of MIRNetUTVNet vs MIRNet · sRGB-SID testPSNR (dB): 22.345 -> 22.691Improved
test UTVNet versus D&E on unseen ELD Canon cameraUTVNet vs D&E · ELD CanonEOS700DPSNR (dB): 16.639 -> 16.496Hurt
test UTVNet versus D&E on ELD Sony and Nikon camerasUTVNet vs D&E · ELD SonyA7S2 and NikonD850PSNR (dB): 17.926 -> 18.161Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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