MIRA: 多模态迭代推理代理用于图像编辑
MIRA: Multimodal Iterative Reasoning Agent for Image Editing
AI总结：
MIRA通过多模态迭代推理代理提升图像编辑的语义一致性和感知质量，结合自研数据集和训练流程，实现与专有系统相当的性能。
摘要：
指导性图像编辑为用户提供了一种通过自然语言编辑图像的直观方式。然而，基于扩散的编辑模型往往难以准确理解复杂的用户指令，尤其是涉及组合关系、上下文线索或指代表达的指令，导致编辑内容语义漂移或未能反映预期的更改。我们通过提出MIRA（多模态迭代推理代理）来解决这个问题，这是一种轻量级、即插即用的多模态推理代理，通过迭代的感知-推理-行动循环进行编辑，有效模拟了多轮人-模型交互过程。不同于发出单一提示或静态计划，MIRA逐步预测原子编辑指令，利用视觉反馈做出决策。我们的15万个多模态工具使用数据集MIRA-Editing，结合两阶段SFT + GRPO训练流程，使MIRA能够处理复杂的编辑指令。当与开源图像编辑模型如Flux.1-Kontext、Step1X-Edit和Qwen-Image-Edit结合时，MIRA显著提升了语义一致性和感知质量，实现了与专有系统如GPT-Image和Nano-Banana相当或更优的性能。
英文摘要：
Instruction-guided image editing offers an intuitive way for users to edit images with natural language. However, diffusion-based editing models often struggle to accurately interpret complex user instructions, especially those involving compositional relationships, contextual cues, or referring expressions, leading to edits that drift semantically or fail to reflect the intended changes. We tackle this problem by proposing MIRA (Multimodal Iterative Reasoning Agent), a lightweight, plug-and-play multimodal reasoning agent that performs editing through an iterative perception-reasoning-action loop, effectively simulating multi-turn human-model interaction processes. Instead of issuing a single prompt or static plan, MIRA predicts atomic edit instructions step by step, using visual feedback to make its decisions. Our 150K multimodal tool-use dataset, MIRA-Editing, combined with a two-stage SFT + GRPO training pipeline, enables MIRA to perform reasoning and editing over complex editing instructions. When paired with open-source image editing models such as Flux.1-Kontext, Step1X-Edit, and Qwen-Image-Edit, MIRA significantly improves both semantic consistency and perceptual quality, achieving performance comparable to or exceeding proprietary systems such as GPT-Image and Nano-Banana.
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。
