AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Fine-grained Image Quality Assessment for Perceptual Image Restoration

Xiangfei Sheng, Xiaofeng Pan, Zhichao Yang, Pengfei Chen, Leida Li

AAAI 2026 · Image generation › Autoregressive and latent generation · approach: Learning-based image quality assessment · 9 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
evaluate within narrow MOS ranges instead of the whole datasetexisting IQA metrics (DeQA-Score) · PIPALSRCC: 0.747 -> 0.350Hurt
evaluate within narrow MOS ranges instead of the whole datasetDB-CNN · PIPALSRCC: 0.636 -> -0.016Hurt
use FGResQ (dual-branch score and ranking) instead of DeQA-ScoreFGResQ · FGRestore (average)pairwise ranking ACC: 0.680 -> 0.752Improved
use FGResQ instead of DeQA-ScoreFGResQ · FGRestore (average)SRCC: 0.678 -> 0.703Improved
use FGResQ instead of full-reference A-FINE for rankingFGResQ · FGRestore (average)pairwise ranking ACC: 0.736 -> 0.752Improved
use FGResQ instead of DeQA-Score on derainingFGResQ · FGRestore DerainingSRCC: 0.507 -> 0.496Hurt
use FGResQ instead of DeQA-Score on super-resolutionFGResQ · FGRestore Super-ResolutionSRCC: 0.561 -> 0.521Hurt
use FGResQ instead of DeQA-Score on dehazingFGResQ · FGRestore DehazingSRCC: 0.762 -> 0.821Improved
pairwise preference ranking in place of scalar MOS for subtle quality differencesIQA annotation · FGRestoreMOS-preference consistencyImproved

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