http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
AI 객체검출 기계학습 시 공포상세도의 첨차 중첩 미인식 사례 연구
전해완(Chun, Hae-Wan),이승재(Lee, Seungjae),류성룡(Ryoo, Seong-Lyong) 대한건축학회 2022 대한건축학회 학술발표대회 논문집 Vol.42 No.2
This study focuses on the AI object detection errors of Chum-Cha parts in Gongpo detail drawing. AI object detection is to detect four main components such as Ju-du, Soro, Sal-mi, and Chum-cha in Gongpo, and its performance and analysis are essential to developing AI-based CAD applications. We found Chum-Cha detection errors in the overlapped drawing parts. The Chum-Cha parts in the Gongpo detail drawing are represented without elevation difference, so AI object detection makes an error. We examine various cases of Chum-Cha parts, and experimental results confirm our analysis.
주심포 공포 AI 객체검출 기계학습시 도면 표현과 부재 인식 과정의 관련성 연구
이준규(Lee, June-Kyu),전해완(Chun, Hae-Wan),이승재(Lee, Seung-Jae),류성룡(Ryoo, Seong-Lyong) 대한건축학회 2022 대한건축학회 학술발표대회 논문집 Vol.42 No.2
The purpose of this study is to analyze the judgment of AI object detection by comparing the method of traditional architecture experts and the results of AI object detection. The results of AI object detection for Gongpo members can be summarized as follows. First, AI object detection understands the structure of Gongpo even if several members are overlapped in the drawing. Second, when AI recognizes Gongpo members, it determines based on the location and relationship of the members, which is not very different from the criteria distinguished by experts. Third, in addition to the location and relationship of the member, the shape of the member is also used, but it is not yet completely distinguished.
전통 목조 건축 교육을 위한 AI CAD 도구 활용 연구 - 주심포 공포를 대상으로 -
이승재(Lee, Seungjae),정다운(Jung, Da-Un),한명기(Han, Myongki),전해완(Chun, Hae-Wan),류성룡(Ryoo, Seong-Lyong) 대한건축학회 2023 대한건축학회 학술발표대회 논문집 Vol.43 No.1
This paper explores the use of AI-based CAD tools for teaching traditional wooden architecture focusing on Gong-po, a key component of vertical expansion and load support. Gong-po consists of several members that are challenging for students to understand in a short time. We developed AI-based CAD tools that enable students to identify Gong-po’s members in drawings and generate 2D and 3D CAD models of them. We also taught a four-week course at a university using these tools. A survey showed that AI-based CAD tools helped students understand Gong-po and its members better.