AI Paper Atlas

Research map / Image generation / Image restoration

Unsupervised Imaging Inverse Problems with Diffusion Distribution Matching

Giacomo Meanti, Thomas Ryckeboer, Michael Arbel, Julien Mairal

ICCV 2025 · Image generation › Image restoration · approach: Diffusion-based universal restoration · 2 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
learn the blur kernel from a dataset by distribution matching instead of from a single image (Blind-DPS)DiffPIR deblurring · FFHQ Gaussian blurPSNR: 29.8 -> 32.7Improved
learn the blur kernel by distribution matching instead of Blind-DPSDiffPIR deblurring · FFHQ motion blurPSNR: 25.5 -> 28.8Improved
use learned kernel instead of true kernelDiffPIR · FFHQ motion blurPSNR: 29.3 -> 28.8Hurt
use learned degradation instead of DeFlowESRGAN · DPDD spatially varying blurPSNR: 27.92 -> 32.38Improved
use learned degradation instead of true kernelDPIR · DPDD spatially varying blurPSNR: 34.89 -> 33.84Hurt
use learned degradation instead of true degradationESRGAN · DPDD spatially varying blurLPIPS: 0.058 -> 0.061No clear effect
use matched-degradation method instead of pretrained RestormerDPDD deblurring · DPDD spatially varying blurPSNR: 28.59 -> 33.84Improved
use learned kernel with USRNet instead of KernelGAN estimatex2 super-resolution · DIV2KRKY-PSNR: 29.65 -> 31.11Improved
plug learned kernel into DANv2 restorer instead of DKPx2 super-resolution · DIV2KRKY-PSNR: 23.18 -> 31.86Improved
plug learned kernel into DANv2 restorer instead of ZSSR (ablation)x2 super-resolution · DIV2KRKY-PSNR: 31.61 -> 31.86Improved
use estimated kernel instead of KernelGANx2 super-resolution · DIV2KRKkernel PSNR: 51.5 -> 64.2Improved
run second stage with 5 random seedsOurs + DiffPIR · FFHQ motion blurPSNR: 0.02No clear effect

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