- 學位論文
人工智慧與擴增實境於裝修工程的應用
Application of Artificial Intelligence and Augmented Reality in interior decoration project management
摘要
進度表為工程專案時程管理之重要依據，為精確掌握專案進度，需經常更新和修改，然而工程專案除了作業項目繁多，作業關係變動頻繁，以人工方式更新修改相當費時、費力，因此許多研究發展出進度自動更新方法。 現有進度自動更新方法主要可以分為兩類：第一類透過Ultra-wideband (UWB)、Radio Frequency Identification (RFID)、Barcode等感測器 (Sensor) 讀取目標物的標籤，通過標籤中的資訊取得工程進度，但因為標籤只能安裝於實體物件，因此只適用於如柱、樑、版及牆等結構體元件外，其他則不適用；第二類應用人工智慧 (Artificial Intelligence) 圖像識別的方式，對工地現場圖像或點雲資料進行“物體偵測”(Object Detection)，通過比對不同時間段的圖像識別結果，紀錄新增的工程進度，但存在辨識率較低的問題。 此外現有方法著重於結構體工程，然而室內裝修工程也是工程專案進度管理的重要一環，又則工程進度並非一元件的數量來衡量，例如一面牆包含鋼筋綁紮、模板、混凝土澆置、粉刷、油漆等不同作業階段，所以辨識牆面的數量並不能用以衡量工程進度，因此本研究以室內裝修工程為例，應用擴增實境技術 (Augmented Reality) 來收集不同工作區域的每天施工進度照片，並透過人工智慧 "圖像分類 " (Image Classification) 來辨識照片的作業階段，進而更新工程進度。
並列摘要
The schedule is an essential basis for project time management. It needs to be updated and modified frequently to accurately monitor the project progress, particularly for projects with many activities, such as interior decoration projects. Updating and revising manually is time-consuming and labour-intensive. Therefore, various automatic schedule update methods are being developed. Current automatic progress update methods can be categorized into two types: one uses ultra-wideband (UWB), radio frequency identification (RFID), barcodes, and other sensors to read the target object's tag. However since tags can only be attached to physical objects, they are only applicable to structural elements, such as columns, beams, slabs, and walls. The second one uses artificial intelligence to detect objects to monitor progress, but the recognition rate is generally low. Additionally, the current methods primarily focus on structural works. However interior decoration is also crucial. Moreover, the number of components completed in a project does not determine its progress. Take the construction of a wall, for example. It involves several stages, including "reinforcement tying", "formwork", "concrete placement" and "painting". Identifying the number of walls cannot be used to determine its progress. This study using interior decoration projects as an example proposes a system that updates the progress of a construction project by determining the construction phase and stages of the works instead of the number of elements. Augmented Reality(AR) technology is used to collect photos of the construction progress each day, and the image classification of Artificial Intelligence(AI) technique is applied to classify the stages of the photos.