文献类型：期刊文献

中文题名：以数据驱动的建筑设计教学改革——基于AI评分系统的全过程融合实践研究

英文题名：Data-driven architectural design teaching reform:a practical study on the whole-process integration of an AI-based scoring system

作者：马庚 齐翔胜 韩会庆

第一作者：马庚

机构：[1]贵州理工学院建筑与城市规划学院,贵州贵阳550025;[2]贵州省黔南州住建局,贵州都匀558000

第一机构：贵州理工学院建筑与城市规划学院

年份：2026

期号：1

起止页码：68-75

中文期刊名：贵州农机化

基金：贵州理工学院教育教学改革研究项目(JGYB202419)。

语种：中文

中文关键词：建筑设计教学;人工智能;自动评分系统;YOLOv8;教育数字化

外文关键词：architectural design teaching;artificial intelligence;automated scoring system;YOLOv8;digitalization of education

摘要：围绕建筑设计教学中存在的评分主观、反馈滞后与资源不足等问题,构建了一套基于深度学习的总平面图AI自动评分系统,融合Open CV、YOLOv8与EasyOCR等算法,实现关键图纸要素的精准识别与结构化评分。将该系统嵌入“建筑设计(三)”课程全过程,并通过58份问卷验证其在提升教学反馈效率、自主优化能力与图纸规范表达方面的有效性。该研究还提出协同教学机制与双维度评价框架,为“新工科”背景下建筑教育智能化改革提供实践范式与技术支撑。
This study addresses key challenges in architectural design teaching,including subjective evaluation,delayed feedback,and limited instructional resources,by developing a deep learning-based AI scoring system for site plan drawings.Integrating algo rithms like OpenCV,YOLOv8,and EasyOCR,this system achieves accurate identification of critical design elements and generate structured evaluations.The system is fully embedded into the Architectural Design(III)course,and its effectiveness in improving teaching feedback efficiency,self-improvement ability,and standardized drawing expression was validated through 58 questionnaires.The study further proposes a collaborative teaching model and a dual-dimensional evaluation framework,offering a practical paradigm and technical support for the intelligent reform of architectural teaching under the"New Engineering"initiative.