From Semantic LiDAR Point Clouds to As-Built BIM: An AI Framework for 3D Reconstruction of Transport Infrastructure

The rapid development of transport infrastructure highlights the need for efficient and accurate methods to reconstruct digital 3D models for asset information management. This study presents an AI-driven framework that integrates semantic light detection and ranging (LiDAR) point clouds with as-built building information modeling (BIM) reconstruction for mass rapid transit (MRT) stations. The proposed approach combines semantic segmentation with a geometry-aware, type-specific instance segmentation strategy, where spatially isolated components are clustered using voxel- and density-based spatial clustering of applications with noise (DBSCAN), while interconnected components are separated through orientation-based grouping and clustering. Furthermore, this study presents a scientific formulation for geometric representation and logical relationships among architectural and mechanical, electrical, and plumbing (MEP) components that bridges semantic instance segmentation to BIM reconstruction. Parametric modeling and matching-based modeling are developed to accommodate components with different characteristics. Experimental results demonstrate high segmentation accuracy (overall accuracy, OA, above 97%; mean intersection over union, mIoU, above 90%) and BIM geometric accuracy (above 80% at a 10 mm geometric deviation threshold), as well as remarkable savings in terms of manpower and time consumption (above 80%).

Authors

Institutions

Publication Details

Journal
Journal of Computing in Civil Engineering
Published
2026-11-21
DOI
https://doi.org/10.1061/jccee5.cpeng-7978
Primary Topic
3D Surveying and Cultural Heritage
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

From Semantic LiDAR Point Clouds to As-Built BIM: An AI Framework for 3D Reconstruction of Transport Infrastructure

Vincent J.L. Gan, Xiuqi Li, Mingkai Li, Lim Pia Lian et al.
Journal of Computing in Civil Engineering
3D Surveying and Cultural Heritage
article

From Semantic LiDAR Point Clouds to As-Built BIM: An AI Framework for 3D Reconstruction of Transport Infrastructure

Vincent J.L. Gan, Xiuqi Li, Mingkai Li, Lim Pia Lian, Kelvin Goh Chun Keong, Allan Yeo Chen Long, Kexin Li
article en

Abstract

The rapid development of transport infrastructure highlights the need for efficient and accurate methods to reconstruct digital 3D models for asset information management. This study presents an AI-driven framework that integrates semantic light detection and ranging (LiDAR) point clouds with as-built building information modeling (BIM) reconstruction for mass rapid transit (MRT) stations. The proposed approach combines semantic segmentation with a geometry-aware, type-specific instance segmentation strategy, where spatially isolated components are clustered using voxel- and density-based spatial clustering of applications with noise (DBSCAN), while interconnected components are separated through orientation-based grouping and clustering. Furthermore, this study presents a scientific formulation for geometric representation and logical relationships among architectural and mechanical, electrical, and plumbing (MEP) components that bridges semantic instance segmentation to BIM reconstruction. Parametric modeling and matching-based modeling are developed to accommodate components with different characteristics. Experimental results demonstrate high segmentation accuracy (overall accuracy, OA, above 97%; mean intersection over union, mIoU, above 90%) and BIM geometric accuracy (above 80% at a 10 mm geometric deviation threshold), as well as remarkable savings in terms of manpower and time consumption (above 80%).

Journal of Computing in Civil EngineeringVol. 41(2)
National University of Singapore (SG), Land Transport Authority (SG)
3D Surveying and Cultural Heritage
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon