发表机构
Fudan University; ByteDance Inc.（复旦大学; 字节跳动公司）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
IterCAD是一种迭代框架，将正交视图转CAD代码任务重构为渐进式程序修复，通过三阶段训练与监督集优化，在CADExpert上显著提升了代码可执行性与几何保真度。
AI 中文摘要
从带尺寸标注的正交工程图生成可执行的参数化CAD代码是一项具有挑战性的任务，需要几何理解、过程推理和精确的数值预测。现有视觉-语言方法通常将该问题表述为单次生成任务，导致模型无法检查中间CAD结果并纠正早期错误，常生成不可执行的代码或几何不一致的输出。本文提出IterCAD，一种迭代框架，将正交视图到CAD的生成重新表述为渐进式程序修复过程。IterCAD不是单次预测最终CAD代码，而是反复分析当前CAD结果、推理其与目标视图的差异，并明确决定是修改代码还是停止优化过程。为使迭代修复可学习，我们进一步构建IterCAD-RS，即包含可修复中间CAD状态和已正确状态的结构化修改-停止监督集，并开发三阶段训练策略，用于初始生成、修改学习和多轮强化学习（RL）优化。通过闭合视觉理解、几何验证和代码优化之间的循环，IterCAD逐步纠正结构和参数错误。在CADExpert数据集上的实验表明，与强大的单次生成基线相比，IterCAD始终提升了代码可执行性和几何保真度。
英文摘要
Generating executable parametric CAD code from dimension-annotated orthographic drawings is a challenging task requiring geometric understanding, procedural reasoning, and precise numerical prediction. Existing vision-language approaches typically formulate this problem as one-shot generation, preventing the model from inspecting intermediate CAD results and correcting early mistakes, often leading to non-executable code or geometrically inconsistent outputs. In this paper, we propose IterCAD, an iterative framework that reformulates orthographic-view-to-CAD generation as a progressive program repair process. Instead of predicting the final CAD code in a single pass, IterCAD repeatedly analyzes the current CAD result, reasons about its discrepancy with the target views, and explicitly decides whether to REVISE the code or STOP the refinement process. To make iterative repair learnable, we further construct IterCAD-RS, a structured revise-or-stop supervision set containing both repairable intermediate CAD states and already-correct states, and develop a three-stage training strategy for initial generation, revision learning, and multi-turn RL optimization. By closing the loop between visual understanding, geometric verification, and code refinement, IterCAD progressively corrects structural and parametric errors. Experiments on CADExpert show that IterCAD consistently improves code executability and geometric fidelity over strong one-shot baselines.
CommentsAccepted to the 34th ACM International Conference on Multimedia (ACM MM 2026)
