论文HF Daily Papers · arXiv 2610.412603 天前

SketchDirector：基于智能体布局推理的复杂场景矢量草图生成框架

提出SketchDirector智能体框架，通过LLM分解贝塞尔曲线资产与自演化记忆循环，实现文本驱动的复杂场景矢量草图生成。配套发布含8K可编辑矢量资产的MCSketch8K数据集，在语义组合质量上超越现有SOTA基线。

AI 整理

仿真与训练
数据与采集
具身模型

本文由 AI 根据公开内容整理，可能有误，请以原文为准：SketchDirector: Complex Scene Sketching via Agentic Layout Reasoning and Detail Refinement（HF Daily Papers · arXiv 2610.41260）

核心内容

针对文本引导复杂场景草图生成难题，研究团队提出SketchDirector智能体框架，模拟人类绘图工作流实现多概念场景自动绘制。该方案突破传统图像引导局限，通过LLM将场景提示词分解为贝塞尔曲线资产，结合自演化记忆进行空间布局规划，并利用扩散先验增强细节同时保留几何结构。实验表明其在语义与组合质量上优于SOTA，并开源MCSketch8K数据集支持后续研究。

技术架构包含三核心模块：LLM负责将文本Prompt拆解为矢量草稿资产；Agentic Reasoning Loop基于自演化记忆迭代规划资产空间排列；Detail Refinement模块利用Diffusion Priors优化线条质量且不破坏拓扑。配套MCSketch8K数据集提供可编辑矢量素材、组合布局及对象级操作标注，解决了复杂场景草图缺乏结构化训练数据的瓶颈，为具身智能任务规划与仿真环境构建提供高精度矢量表征基础。

关键看点

智能体布局推理：采用Agentic Reasoning Loop结合自演化记忆机制，迭代规划多概念资产的空间排列，解决复杂场景组合逻辑问题
矢量资产表征：利用LLM将场景文本分解为贝塞尔曲线集合，生成可编辑矢量草稿，确保几何结构在细化过程中保持拓扑一致性
专用数据集发布：推出MCSketch8K数据集，包含可编辑矢量资产、组合布局及对象级操作标注，填补多概念场景草图结构化数据空白
扩散先验细化：Detail Refinement模块引入Diffusion Priors增强草图细节表现力，在提升视觉质量的同时有效保留整体轮廓组织

工程与产业观察

该框架将非结构化文本转化为结构化矢量草图的能力，对具身智能任务规划与仿真场景生成具有关键价值。贝塞尔曲线表征与对象级操作标注直接适配机器人运动原语，MCSketch8K数据集可作为Sim-to-Real中语义理解与操作策略学习的中间态数据源，降低复杂环境建模成本。

英文摘要 (Abstract)

Existing sketch generation methods often rely on image guidance and are largely limited to individual objects or simplified scenes, leaving text-guided complex scene sketching underexplored. We present SketchDirector, an agentic framework that mimics the human sketching workflow for automatic multi-concept scene sketching. Given a scene-level text prompt, SketchDirector first employs a large language model~(LLM) to decompose the scene into multiple draft assets, where each is represented by a set of Bézier curves. Then, an agentic reasoning loop iteratively plans and executes their spatial arrangement, guided by self-evolving memory. Finally, a detail refinement module leverages diffusion priors to enhance sketch quality while preserving geometric structure. To support future research and applications in multi-concept scene sketch generation, we further introduce the MCSketch8K dataset, comprising editable vector assets, compositional layouts, and object-level manipulation annotations. Extensive experiments show that SketchDirector outperforms state-of-the-art baselines in semantic and compositional quality while effectively preserving overall contour organization during refinement. The project page is: https://yiki77.github.io/SketchDirector/{https://yiki77.github.io/SketchDirector/}.

Hugging Face Daily Papers 热度：1 upvotes 作者：Yiqi Liang, Xiaodong Cun