Building information modelling of staircase

• Main Author: Li, Ji an
• Other Authors: Cai Yiyu
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Mechanical engineering; Engineering::Industrial engineering
• Online Access: https://hdl.handle.net/10356/167721
• Institution: Nanyang Technological University

With the development of information technology, artificial intelligence, big data, and Internet of Things, BIM (Building Information Modelling) has brought a significant digital transformation in the architecture, engineering, and construction (AEC) industry. Thanks to the efficient scanning technology LiDAR and accurate representation of the spatial position of the point cloud, 3D models of large buildings, pipelines, room structures, and other scenes that require a large number of engineering measurements and manual drawings can be reconstructed by Scan-to-BIM techniques. This report focuses on automatic reconstruction of staircase BIM model from point cloud, aiming to improve the reconstruction accuracy based on Region Growing segmentation by adding deep learning segmentation labels. Inspired by the Point Transformer model, which utilizes self-attention networks for 3D point cloud processing, we fine-tuned a staircase segmentation model using simulation data and a pre-trained indoor semantic segmentation model. By adding deep learning segmentation labels, the Building Information Modelling accuracy improves and the process becomes universal for various staircase dimensions. Keywords: Scan-to-BIM, As-Built BIM Model, Point cloud planar segmentation, Deep Learning sematic segmentation