EditRefiner：一种对齐人类的代理框架用于图像编辑细化
EditRefiner: A Human-Aligned Agentic Framework for Image Editing Refinement
AI总结：
本文提出EditRefiner，一种对齐人类的代理框架，通过人类反馈数据集改进图像编辑的细粒度问题，通过感知-推理-行动-评估循环提升编辑质量。
摘要：
最近的文本引导图像编辑（TIE）模型取得了显著进展，但编辑后的图像仍经常出现细粒度问题，如不自然的对象、光照不匹配和意外变化。现有的细化方法要么依赖于昂贵的迭代再生，要么使用视觉语言模型（VLMs）具有弱空间定位，往往导致语义漂移和不可靠的局部修正。为了解决这些限制，我们首先构建了EditFHF-15K数据集，包含15K张来自12个TIE模型的图像，涵盖43种编辑任务，60K个标注的瑕疵区域和80K个编辑失败区域，每个区域都配有文本推理，以及45K个平均意见分数（MOSs）评估感知质量、指令遵循和视觉一致性。基于EditFHF-15K，我们提出了EditRefiner，一种分层、可解释且对齐人类的代理框架，将后编辑修正重新表述为人类般的感知-推理-行动-评估循环。具体来说，我们引入：（1）一个感知代理，检测瑕疵和编辑失败的上下文显著图；（2）一个推理代理，解释这些感知线索以执行对齐人类的诊断推理；（3）一个行动代理，使用推理输出计划和执行局部重编辑；（4）一个评估代理，评估重编辑的图像并指导行动代理是否需要进一步细化。广泛的实验表明，EditRefiner在畸变定位、诊断准确率和人类感知对齐方面始终优于最先进的方法，建立了自我纠正和感知可靠的图像编辑新范式。代码可在https://github.com/IntMeGroup/EditRefiner获取。
英文摘要：
Recent text-guided image editing (TIE) models have made remarkable progress, yet edited images still frequently suffer from fine-grained issues such as unnatural objects, lighting mismatch, and unexpected changes. Existing refinement approaches either rely on costly iterative regeneration or employ vision-language models (VLMs) with weak spatial grounding, often resulting in semantic drift and unreliable local corrections. To address these limitations, we first construct EditFHF-15K, a dataset of fine-grained human feedback for edited images, comprising (1) 15K images from 12 TIE models spanning 43 editing tasks, (2) 60K annotated artifact regions and 80K editing failure regions, each accompanied by textual reasoning, and (3) 45K mean opinion scores (MOSs) assessing perceptual quality, instruction following, and visual consistency. Based on EditFHF-15K, we propose EditRefiner, a hierarchical, interpretable, and human-aligned agentic framework that reformulates post-editing correction as a human-like perception-reasoning-action-evaluation loop. Specifically, we introduce: (1) a perception agent that detects contextual saliency maps of artifacts and editing failures, (2) a reasoning agent that interprets these perceptual cues to perform human-aligned diagnostic inference, (3) an action agent that uses the reasoning output to plan and execute localized re-editing, and (4) an evaluation agent that assesses the re-edited image and guides the action agent on whether further refinements are required. Extensive experiments demonstrate that EditRefiner consistently outperforms state-of-the-art methods in distortion localization, diagnose accuracy and human perception alignment, establishing a new paradigm for self-corrective and perceptually reliable image editing. The code is available at https://github.com/IntMeGroup/EditRefiner.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
