AI Paper Atlas

Research map / Image generation / Image restoration

MMDIR: Multimodal Instruction-Driven Framework for Mixed-Degradation Document Image Restoration

Heng Li, Xingyuan Wang, Yang Fan, Yunan Zhang, Xiangping Wu, Qingcai Chen

CVPR 2026 · Image generation › Image restoration · approach: Efficient restoration architectures

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 LLM-based text instruction and degradation identification (ablation)MMDIR · BMVC deblurringPSNR: 26.76 -> 29.03Improved
add LLM-based text instruction and degradation identification (ablation)MMDIR · OSR deshadowPSNR: 18.79 -> 19.60Improved
add LLM-based text instruction and degradation identification (ablation)MMDIR · MixedDocPSNR: 23.83 -> 24.43Improved
add text instruction (ablation)MMDIR · MixedDocLPIPS: 0.1324 -> 0.1217Improved
add text instruction (ablation)MMDIR · BMVC deblurringDISTS: 0.0320 -> 0.0233Improved
use MMDIR instead of retrained DocDiff and DiffUIR on mixed degradationsMMDIR · MixedDocPSNR: 22.09 (DiffUIR1024*) -> 24.43Improved
use MMDIR instead of retrained FoundIR on mixed degradationsMMDIR · MixedDocFID: 10.41 -> 9.07Improved
use MMDIR for downstream OCR on restored imagesMMDIR · MixedDocCER: 0.23 (FoundIR*) -> 0.17Improved
use MMDIR vs LGA-Doc on deblurring SSIMMMDIR · BMVC deblurringSSIM: 0.978 (LGA-Doc) vs 0.977No clear effect
use MMDIR vs BGSNet and DiffUIR on deshadow PSNRMMDIR · OSR deshadowPSNR: 21.25 (BGSNet) vs 19.60Hurt
increase number of co-occurring degradation types from one to fourMMDIR · MixedDocPSNR and SSIMHurt
increase number of degradation types per imageMMDIR degradation identification · MixedDocrecall: 1 (1 type) -> .9647 (4 types)Hurt

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