Publications / 2022 Proceedings of the 39th ISARC, Bogotá, Colombia

Automated Checking of Scaffold Safety Regulations using Multi-Class 3D Segmentation

Jeehoon Kim, Juhyeon Kim, Nahye Koo and Hyoungkwan Kim
Pages 115-119 (2022 Proceedings of the 39th ISARC, Bogotá, Colombia, ISBN 978-952-69524-2-0, ISSN 2413-5844)

Scaffolds, one of the most widely used temporary structures, are prone to safety-related accidents. Despite the fact, checking regulations for a scaffold is manually being conducted, which is inefficient, especially for a large construction site. This paper proposes an automated method to check safety regulations regarding scaffolds on sites. 3D point cloud data obtained from Terrestrial Laser Scanning (TLS) is first processed by a deep learning-based 3D segmentation to automatically identify major entities Then, a simple rule-based algorithm is applied to the segmented data to check for three types of major safety-related regulations. The result of our experiment shows potential for successfully automating scaffold safety checking at a construction site.

Keywords: Deep learning; Scaffold; Point cloud; Semantic Segmentation; Safety regulation checking; Terrestrial Laser Scanning (TLS)