隨著人工智慧的蓬勃發展，深度學習技術的應用越來越多元。本研究利用影像分割技術對取得的工地影像進行影像的分析處理，並開發一系統整合影像分割模型，利用單一攝影機進行大範圍施工進度控管，辨識六種常見施工階段，幫助管理人員更有效率的掌握工程進度。
本研究應用影像辨識中的深度學習物件分割技術，通過攝影機和網路獲取的施工圖片，以大範圍施工之照片為蒐集方向建立工地常見六種工項的數據集，並使用遷移式學習進行訓練。將工地現場攝影機的影像與施工工地的BIM模型底圖視角進行套疊，通過本研究系統，可即時對工地影像進行物件分割，並與視角套疊後之模型進行比對處理，從而實現工地進度的即時自動化控管。
在工地影像經過分割並與相同角度的模型進行比對的過程中，系統處理影像與模型的像素，使像素匹配後進行比對。分割後的工地影像在相同視角模型中的像素面積百分比，即為該工項的施工進度。

This study employs image segmentation to analyze and process construction site images and develops a system integrating segmentation models. Using a single camera for large-scale construction monitoring, it identifies six common construction stages, aiding managers in efficiently tracking project progress.
The study applies deep learning object segmentation for image recognition.It establishes a dataset of six common construction items from extensive photos and uses transfer learning for training. Site images are overlaid with the BIM model perspective, enabling real-time object segmentation and comparison, thus achieving automated construction progress control.
During the comparison of segmented site images with models from the same angle, the system processes the pixels of the images and models, matching pixels for comparison. The percentage of pixel area of the segmented construction site image in the model from the same perspective represents the construction progress of the specific construction item.

[1] 陳宛榆(2023),應用深度學習技術於施工現場影像之工項進度自動化偵測,國立臺灣科技大學碩士學位論文
[2] 吳庭瑜(2023), 整合AI影像辨識與BIM技術之自動化施工進度管控系統,國立臺灣科技大學碩士學位論文
[3] 趙宏斌(2022), 深度學習在土木工程管理的應用
[4]Qurishee, M.;Wu, W.;Atolagbe, B.;Owino, J.;Fomunung, Creating a Dataset to Boost Civil Engineering Deep Learning Research and Application,Mar.2020,Vol.12 No.3
[5]Ghorbanzadeh, O.; Blaschke, T.; Gholamnia, K.; Meena, S.R.; Tiede, D.; Aryal, J. Evaluation of Different Machine Learning Methods and Deep-Learning Convolutional Neural Networks for Landslide Detection. Remote Sens. Nov.2019,No.196
[6] Jonathan Long;Evan Shelhamer;Trevor Darrell,Fully convolutional networks for semantic segmentation,Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015, pp. 3431-3440
[7] Ghosh ;Swarnendu ;Das Nibaran ;Ishita ;Maulik ;Ujjwal,Understanding Deep Learning Techniques for Image Segmentation, July.2019, Vol.52 No.4
[8] Shijie Hao;Yuan Zhou;Yanrong Guo, A Brief Survey on Semantic Segmentation with Deep Learning,Nov.2019,Vol.406,pp. 302-321
[9] M. Maalej ; R. Cherif ; Y. Yaddaden ;A. Khoumsi,Automatic Crack Detection on Concrete Structure Using a Deep Convolutional Neural Network and Transfer Learning,2022 2nd International Conference on Advanced Electrical Engineering (ICAEE),Oct. 2022, pp. 1-6
[10] Yuanzhi Xiang ;Jiayue Zhao ;Wenjing Wu ; Automatic Object Detection of Construction Workers and Machinery Based on Improved YOLOv5,Sep.2022
[11] Hoskere, V.;Narazaki, Y.; Hoang, T.A; MaDnet: multi-task semantic segmentation of multiple types of structural materials and damage in images of civil infrastructure,J Civil Struct Health Monit,May 2020,pp.757–773
[12] Zhuang, F.; Qi, Z.;Duan, K.;Xi, D.;Zhu, Y.;Zhu, H.; Xiong, H.;& He, Q.; A Comprehensive Survey on Transfer Learning. ,Nov. 2019. Proceedings of the IEEE, 109, pp. 43-76.
[13] Di, X.; Li, L.; Image Defect Recognition Method Based on Deep Learning Network,Mar. 2022. Mobile Information Systems
[14] Gerace, F.;Doimo, D.;Mannelli, S.;Saglietti, L.;& Laio, A, Optimal transfer protocol by incremental layer defrosting,Mar.2023.
[15]Mask R-CNN模型架構圖,
https://github.com/huaweicloud/ModelArts-Lab/wiki/Mask-R-CNN
[16] Liang-Chieh Chen;George Papandreou;Florian Schroff;Hartwig Adam,
Rethinking Atrous Convolution for Semantic Image Segmentation, Dec. 2017
[17] Shu Liu; Lu Qi; Haifang Qin; Jianping Shi; Jiaya Jia,Path Aggregation Network for Instance Segmentation, Sep. 2018
[18] YOLOv8物件分割模型綜合效能指標, https://github.com/ultralytics/ultralytics
[19] YOLOv8模型架構圖,
http://www.tcsae.org/supplement
[20]PyQt,[online],https://pypi.org/project/PyQt5/
[21]Riverbank Computing,[online],https://riverbankcomputing.com/
[22]Unity,[online], https://unity.com/
[23]謝政廷(2021),整合SLAM-based 頭戴式MR與BIM之室內施工進度視覺化管理遠端協作系統,國立臺灣科技大學碩士學位論文
[24]陳品至(2023),整合頭戴式混合實境與AI影像辨識之施工機具安全檢核系統,國立臺灣科技大學碩士學位論文
[25]WebRTC, [online],https://webrtc.org/?hl=zh-tw