Publications / 2024 Proceedings of the 41st ISARC, Lille, France

Automated Decision Support System Based on Quantification of Defective Tubular Steel Temporary Materials for Quality Circles

James Mugo Njoroge, Seojoon Lee, Kyuhyup Lee, Junsung Seol, Younghee Chang, Soonwook Kwon
Pages 888-895 (2024 Proceedings of the 41st ISARC, Lille, France, ISBN 978-0-6458322-1-1, ISSN 2413-5844)
Abstract:

Temporary materials are stored in construction warehouses after and before they are used in construction sites. Most of these materials are made of steel tubular sections such as square, circular, and rectangular sections. However, due to frequent reuse and their storage conditions they become subject to defects such as rust and bend which affect their quality for use in construction projects. Since these materials are stored in stacks or batches, checking the materials individually for defects can be time consuming making end point surface defect detection more efficient for construction workers. Deep learning techniques have proven to be more efficient than manual inspection. However, quantification of materials with defects for decision making on reusing, repairing, and disposing actions and documentation is still a challenge for construction workers. Hence, this paper quantifies temporary tubular steel materials; square hollow section, circular hollow section and rectangular hollow section with common cross section area defects using deep learning technique connected to a web platform for construction quality circle workers. The quantified defective materials are then used for decision making on the actions to be undertaken using statistical distribution with a reporting and documentation option using the web framework. Spatial pyramid pooling layer with cross stage partial network was selected due to higher computation of 105.2 GFLOPS and mAP of 84.9 for the deep learning model. Small and normal batches recorded high accuracy with an average inference speed of 20 seconds per inference

Keywords: Deep Learning, Decision Support System, Temporary Materials, Defects, Quality Circles.