在台灣建築設計、土木工程與施工營造（Architecture,Engineering,Construction, AEC）產業中，室內格局初步規劃仍仰賴人工丈量與圖說繪製，然而人工丈量費時費力，尤其在面積較大或構造複雜的空間中，常需數日才能完成初步測繪，進而影響整體設計進度；此外，室內空間常見的家具與雜物遮蔽亦可能導致丈量資料不完整或誤差，增加設計與實際狀況不符的風險，儘管市面已有多款手機 AR 量測工具，仍受限於鏡頭品質與現場環境，對複雜室內形狀的應用成效有限。
隨著3D掃描與人工智慧技術發展日趨成熟，點雲資料逐漸廣泛應用於設計場域，本研究提出一套整合智慧型手機拍攝影片與神經輻射場（NeRF）重建演算法之流程，有效重建室內三維幾何資訊，加速空間數位丈量與現況圖面產製。
本研究首先透過專家訪談與文獻整理，歸納設計流程中的操作需求與常見痛點，並以 IDEF0 建立涵蓋拍攝、點雲分類至圖面生成的完整作業架構，以符合室內設計對於高效率、低重工風險與良好業主溝通之需求，實作層面上，本研究應用PointNet++模型自動分類點雲中門窗、牆柱等構件，並結合Python腳本與ezdxf工具，產製封閉空間輪廓之2D現況圖面，支援初稿繪製等實務需求，最後透過虛擬現況圖面尺寸與現地丈量數據之幾何比對，驗證所建構流程在空間邊界與室內高度之精準度，證實本研究提出之自動化作業符合室內設計需求。

In Taiwan, interior layout planning still relies on manual measurements and 2D drafting. However, such measurement tasks are labor-intensive and time-consuming, particularly in large-scale or complex spaces where designers often require several days to complete the surveying process, leading to overall project delays. Moreover, obstacles such as existing furniture and objects impede accurate measurement, resulting in incomplete or erroneous data and increasing the risk of design discrepancies. Although various smartphone-based AR measurement applications are available, their accuracy remains constrained by environmental conditions, making them less effective for complex indoor geometries. With the rapid improvement of 3D scanning and artificial intelligence technologies, point cloud data has been increasingly adopted in the field of interior design. By leveraging video captured with a smartphone in conjunction with Neural Radiance Fields (NeRF) reconstruction algorithms, highly accurate 3D geometric information of indoor spaces can be efficiently generated, accelerating the digital progression.
This study proposes a systematic workflow for interior layout modeling. It begins with expert interviews and literature review to analyze the practical requirements and challenges of interior design, followed by the development of a comprehensive workflow using the IDEF0 methodology—from video capture and point cloud generation to object classification and interior design drawing. A PointNet++ deep learning model is subsequently trained for automated 3D point cloud classification, focusing on key architectural elements such as doors, windows, walls, and columns. Finally, using Python and the ezdxf library, the classified point cloud data is transformed into precise 2D floor plans and sectional drawings. The generated geometric dimensions are validated against actual measurements to assess the feasibility and accuracy of the proposed system, ultimately enhancing the efficiency and precision of interior design drafting.

英文文獻
[1]FIPS, Integration Definition for Function Modeling (IDEF0), Federal Information Processing Standards, USA, 1993.
[2]Charles R. Qi., Su, H., Mo, K., & Guibas, L. J., PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation, Stanford University , 2017.
[3]Charles R. Qi., Yi, L., Su, H., & Guibas, L. J., PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space, Stanford University, USA, 2017.
[4]Qian-Yi Zhou, Jaesik Park, & Vladlen Koltun., Open3D: A Modern Library for 3D Data Processing, Intel Labs, USA, 2018.
[5]Charles R. Qi, Or Litany, Kaiming He, Leonidas J. Guibas, Deep Hough Voting for 3D Object Detection in Point Clouds, Facebook AI Research and Stanford University, USA, 2019.
[6]Chen, Z., & Wang, X., Application of AI technology in interior design, Department of Architecture and Art, of North China University of Technology, China, 2020.
[7]Jianwen Xie, Yifei Xu, Zilong Zheng, Song-Chun Zhu, Ying Nian Wu, Generative PointNet: Deep Energy-Based Learning on Unordered Point Sets for 3D Generation, Research Paper, UCLA, USA, 2021.
[8]Wu, X., Liu, Y., Tan, J., & Zhang, S., Automatic Semantic Segmentation for BIM from 3D Point Clouds, Conference Paper, China, 2021.
[9]Wei, Y., Liu, S., Rao, Y., Zhao, W., Lu, J., & Zhou, J., NerfingMVS: Guided Optimization of Neural Radiance Fields for Indoor Multi-view Stereo, Department of Automation, Tsinghua University, China, 2021.
[10]Laurențiu, M. A., Alexandru, I. P., & Lucian, C. R., A Web-Based Platform for 3D Visualization of Multimodal Imaging Data in Cultural Heritage Asset Documentation, National Institute of Research and Development for Optoelectronics INOE 2000, Romania, 2022.
[11]Michael C. P. S., Sophie, Y.Y. L., Ken H.C. C., Henry J. L., Richard H., Scan-to-BIM Technique in Building Maintenance Projects: Practicing Quantity Take-off, School of Architecture and Built Environment, University of Newcastle, Australia, Conference Paper, 2022.
[12]Allie K. Miller, The Reality of AI Product Management, YouTube Video, USA, 2022. https://www.youtube.com/watch?v=eiLudzhvIcs&ab_channel=AllieKMiller
[13]Tan, Y., Chen, L., Wang, Q., Li, S., Deng, T., & Tang, D.D., Geometric Quality Assessment of Prefabricated Steel Box Girder Components Using 3D Laser Scanning and Building Information Model, Shenzhen University, China, 2023.
[14]Ruilong Li, Brent Yi, Justin Kerr, Terrance Wang, Alexander Kolesnikov, Jake An, Kamyar Salahi, Abhik Ahuja, David McAllister, Angjoo Kanazawa, Nerfstudio: A Modular Framework for Neural Radiance Field Development, Conference Papers, University of California, Berkeley, USA, 2023.
[15]Gilad Baruch, Zhuoyuan Chen, Afshin Dehghan, Tal Dimry, Yuri Feigin, Peter Fu, Thomas Gebauer, Brandon Joffe, Daniel Kurz, Arik Schwartz, Elad Shulman., ARKitScenes: A Diverse Real-World Dataset For 3D Indoor Scene Understanding Using Mobile RGB-D Data, Apple Inc., USA, 2023. https://github.com/apple/ARKitScenes
[16]Aurellius, A. R., The Framework of Creating and Searching 3D Objects in the VR Platform: A Case Study of Interior Design, Master’s Thesis, Department of Civil Engineering, National Cheng Kung University, Taiwan, 2023.
[17]Chandan Yeshwanth, Yueh-Cheng Liu, Matthias Nießner, Angela Dai, ScanNet++: A High-Fidelity Dataset of 3D Indoor Scenes, Conference Papers, Technical University of Munich, German, 2023. https://kaldir.vc.in.tum.de/scannetpp/
[18]Rucha Shende, Heritage3DMtl: A Multi-modal UAV Dataset of Heritage Buildings for Digital Preservation, Master's Thesis, Concordia University, Canada, 2024.
[19]Straub, J., Daniel, D., Tianwei, S., Nan, Y., Chris, S., & Richard N., EFM3D: A Benchmark for Measuring Progress Towards 3D Egocentric Foundation Models, Meta Reality Labs Research, USA, 2024.
[20]Siwen, Q., Junhao, Y., Ziming, N., Muze, W., Sijia, F., Pei, A., & Jiaqi, Y.,D., Deep Learning for 3D Point Cloud Enhancement: A Survey, Chang’an University, China, 2024.
[21]Chengyuan Li, Tianyu Zhang, Xusheng Du, Ye Zhang, Haoran Xie, Generative AI for Architectural Design, Tianjin University, 2024.
[22]Abramov, N. et al., Implementing Immersive Worlds for Metaverse-Based Participatory Design through Photogrammetry and Blockchain, Research Paper, Politecnico di Milano, Italy, 2024.
[23]Mostafa Mahmoud, Wu Chen, Yang Yang, & Yaxin Li. Automated BIM generation for large-scale indoor complex environments based on deep learning, Research Paper, The Hong Kong Polytechnic University, China
[24]Jon Stephens, Nerfstudio Gaussian Splatting Implementation, GitHub Repository, USA, 2024. https://github.com/jonstephens85/nerfstudio_guassians
[25]Jon Stephens, How to Install Nerfstudio , YouTube Video, 2024. https://www.youtube.com/watch?v=3JIpZd5XNAc&t=386s&ab_channel=PixelReconstruct
[26]Luma AI, https://lumalabs.ai/dashboard/captures
中文文獻
[1]林耿帆，「以物件為基礎之光達點雲分類」，碩士論文，國立臺灣大學土木工程系，台灣，2012年。
[2]王柏仁，「結合知識本體與BIM於機電設施維護之應用－以大型設備更換為例」，碩士論文，國立臺灣大學土木工程系，台灣，2016年。
[3]張浩彥，「BIM於建築施工流程整合之應用研究」，碩士論文，國立中央大學土木工程系，台灣，2017年。
[4]林琳，〈VR技術在建築室內設計中的應用探討〉，專題報告，遼寧城市建設職業技術學院，中國，2018年。
[5]徐夢茹，〈虛擬現實技術在室內設計中的應用〉，《電腦知識與技術》期刊論文，中國，2018年。
[6]黃祥翔，「建築資訊模型（BIM）整合管理平台開發與效益分析」，碩士論文，國立中央大學土木工程學系，台灣，2019年。
[7]羅塔南，「應用擴增實境增進機電系統之預防性的維護作業」，碩士論文，國立臺灣科技大學建築系，台灣，2021年。
[8]周詠鈞，「結合BIM與光達點雲資料建構符合業主需求之工程進度評量模式—以裝修工程為例」，碩士論文，國立成功大學土木工程系，台灣，2021年。
[9]吳怡潔，董志明，薛帆，「基於LiDAR點雲的竣工BIM建模研究綜述」，第七屆全國BIM學術會議論文，重慶，中國，2021年。
[10]王柔心，「以元宇宙概念建構營建協作平台—以室內裝修為例」，碩士論文，國立成功大學土木工程系，台灣，2022年。
[11]林之謙、曾惠斌、曾仁杰，「應用建築資訊建模（BIM）、深度學習及自動辨識技術輔助建築構件精準安裝」，內政部建築研究所研究報告，國立臺灣大學土木系，台灣，2022年。
[12]龔智群，「人工智慧輔助BIM建築設計流程」，碩士論文，國立中興大學土木工程學系，台灣，2023年。
[13]吳政翊、蔡佳穎，「應用生成式AI於虛擬室內設計—以廚具配置為例」，碩士論文，國立成功大學建築系，台灣，2023年。
[14]王怡慧、王世元、李岳衡等，「以3D點雲掃描技術建構BIM圖資流程」，技術報告，國立臺灣大學土木系，台灣，2023年。
[15]郭頤潔，「利用Omniverse發展虛擬室內裝修平台架構」，碩士論文，國立成功大學土木工程系，台灣，2024年。
[16]行政院農業部水土保持局，〈應用3D Gaussian Splatting法進行地球任何角落無限制範圍的AI手機建模介紹〉，《水保電子報》，台灣，2024年。https://tech.ardswc.gov.tw/EPaper/Home/EPaper?PaperID=fa859c83-f681-43cc-b454-f06da007d621