This technique combines deep learning and digital twin technologies to connect the design and construction process and to achieve construction-site rebar inspection, issue tracking, inspection verification and maintenance. Using image data collection and digital twin creation, cooperative machine learning, and rebar inspection module development as the cores, construction quality inspection will be achieved through matching and comparing between BIM (Building Information Modeling) model and 4D digital twin.
Our breakthrough contains four major parts. First, a dataset of on-site rebar assembly images with high quality labeling and spatial information . Second is the rebar feature recognition model based on a deep learning algorithm. The third is the combination of deep learning and digital twin to realize the point cloud feature recognition of rebars. The fourth is the design-based on-site rebar inspection method, which automatically compares BIM and the point cloud segmentation.

線上展網址：
https://tievirtual.twtm.com.tw/iframe/e40a4362-f76b-4db6-9505-59347d275140?group=23bfb1fa-dd5b-4836-81a1-4a1809b1bae5&lang=en