PulseAugur
EN
LIVE 19:12:04

RePlan framework enhances complex image editing with reasoning and region planning

Researchers have introduced RePlan, a novel framework for instruction-based image editing designed to handle complex instructions and scenes. RePlan employs a reasoning-guided approach where a planner first decomposes intricate instructions and identifies target regions, followed by a diffusion editor that applies changes without iterative inpainting. The system utilizes GRPO-based reinforcement learning for improved reasoning and format reliability, and a new benchmark, IV-Edit, has been developed to evaluate fine-grained grounding and knowledge-intensive edits. RePlan demonstrates superior performance over existing methods in complex scenarios, achieving greater regional precision and overall consistency. AI

IMPACT Enhances AI capabilities in precise, complex image manipulation, potentially improving creative tools and user interfaces.

RANK_REASON The cluster contains a research paper detailing a new method and benchmark for image editing. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

RePlan framework enhances complex image editing with reasoning and region planning

COVERAGE [1]

  1. arXiv cs.CV TIER_1 English(EN) · Tianyuan Qu, Lei Ke, Xiaohang Zhan, Longxiang Tang, Yuqi Liu, Bohao Peng, Bei Yu, Dong Yu, Jiaya Jia ·

    RePlan: Reasoning-guided Region Planning for Complex Instruction-based Image Editing

    arXiv:2512.16864v2 Announce Type: replace Abstract: Instruction-based image editing enables natural-language control over visual modifications, yet existing models falter under Instruction-Visual Complexity (IV-Complexity), where intricate instructions meet cluttered or ambiguous…