Validation of a methodology to automatically classify design stage 3D models into activity units using PointNet and PointNet++ algorithms

A design model comprises the elemental units of a structure. These elements should be extracted and classified into detailed components to use this model during the construction life cycle. However, owing to the increasing scale and complexity of construction projects, the complexity and level of de...

Full description

Saved in:
Bibliographic Details
Main Authors: Jae Hee Lee, Leen Seok Kang
Format: Article
Language:English
Published: Taylor & Francis Group 2025-07-01
Series:Journal of Asian Architecture and Building Engineering
Subjects:
Online Access:http://dx.doi.org/10.1080/13467581.2025.2533231
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A design model comprises the elemental units of a structure. These elements should be extracted and classified into detailed components to use this model during the construction life cycle. However, owing to the increasing scale and complexity of construction projects, the complexity and level of detail of design models have increased significantly due to the growing scale of construction projects, along with the number of components they contain. Therefore, an efficient method to accurately classify the detailed components of the design model should be devised. We used PointNet and PointNet++ to automatically classify construction activity units of bridge design models. The accuracies of these methods for the automatic classification of abutment, pier, and bridge design models were validated, reaching 0.9 ~ 1.0, 1.0, and 0.69 ~ 0.99, respectively. The proposed automatic classification approach may enable the reuse of design models throughout the construction life cycle.
ISSN:1347-2852