FlowTool: Controlling Tool Parameter in Image Retouching via Flow Matching

• 类型：arxiv
• 标识：2609.35673
• 链接：https://arxiv.org/abs/2609.35673
• 主分类：multimodal
• 形态：method
• TLDR：Tool-based image editing (image retouching) is commonly formulated with autoregressive multimodal large language models (MLLMs) that sequentially generate reasoning, tool selections, and parameter values. In this work, we present a novel approach to tool-based image editing by framing the task as a flow matching problem. We introduce FlowTool, a framework that directly models the distribution of high-quality tool parameters conditioned on the input image and user instruction using conditional rectified flow. FlowTool combines a vision-language model backbone for multimodal understanding with a
• 待LLM分类：否
• 标题中文：FlowTool: 基于流匹配控制图像修图中的工具参数
• TLDR中文：基于工具的图像编辑（图像修图）通常被建模为自回归多模态大语言模型（MLLM），依次生成推理、工具选择与参数值。本文将任务重新建模为流匹配问题，提出 FlowTool：使用条件 rectified flow 直接建模高质量工具参数分布（以输入图像与用户指令为条件）。FlowTool 将用于多模态理解的视觉-语言模型主干与一个用于参数生成的流匹配头相结合。
• 来源文件：
• /inbox/tom/_candidates/2026-09-29-agent-rag-longcontext-candidates.json