AI Paper Atlas

Research map / Image generation / Editing and style transfer

RefEdit: A Benchmark and Method for Improving Instruction-based Image Editing Model on Referring Expressions

Bimsara Pathiraja, Maitreya Patel, Shivam Singh, Yezhou Yang, Chitta Baral

ICCV 2025 · Image generation › Editing and style transfer · approach: Instruction-based diffusion editing · 10 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 20000 RefEdit-Data samples to UltraEdit (RefEdit-SD3)UltraEdit · RefEdit-Bench Easyoverall VIEScore: 3.4 -> 4.99Improved
add 20000 RefEdit-Data samples to UltraEdit (RefEdit-SD3)UltraEdit · RefEdit-Bench Hardoverall VIEScore: 2.59 -> 3.98Improved
finetune on RefEdit-Data plus MagicBrush data (RefEdit) instead of MagicBrush aloneInstructPix2Pix SD1.5 · RefEdit-Bench Easyoverall VIEScore: 3.67 -> 4.68Improved
finetune on RefEdit-Data plus MagicBrush data (RefEdit) instead of MagicBrush aloneInstructPix2Pix SD1.5 · RefEdit-Bench Hardoverall VIEScore: 3.56 -> 3.93Improved
use 20000 targeted synthetic samples rather than 1.2M OmniEdit samplesSD1.5 editing model · PIE-Benchoverall VIEScore: 5.12 -> 5.21Improved
use larger base model (SDXL CosXLEdit) instead of SD1.5 MagicBrushimage editing baselines · RefEdit-Bench Hardoverall VIEScore: 3.56 -> 2.88Hurt
use larger base model (SD3 UltraEdit) instead of SD1.5 MagicBrushimage editing baselines · RefEdit-Bench Hardoverall VIEScore: 3.56 -> 2.59Hurt
use SD3 variant (RefEdit-SD3) instead of SD1.5 RefEditRefEdit · PIE-Benchoverall VIEScore: 5.21 -> 5.70Improved
evaluate on RefEdit-Bench Hard instead of PIE-BenchRefEdit-SD3 · RefEdit vs PIE benchmarksoverall VIEScore: 5.70 -> 3.98Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 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