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

3D reconstruction of a bridge with concrete damage classification using deep learning

Christopher Joseph Núñez Varillas, Marck Steewar Regalado Espinoza, Luis Mario Huaypar Acurio, Antonio Stefano Bedon Rosario, Jordan Antony Romaní Chavez, Oscar Manuel Solis Garcia, Karol Maricruz Agreda Estela, Micaela Anthoaneth Cardenas Contreras
Pages 722-729 (2024 Proceedings of the 41st ISARC, Lille, France, ISBN 978-0-6458322-1-1, ISSN 2413-5844)
Abstract:

The classification of concrete damage in bridges poses challenges, characterized by time-consuming, hazardous, and often subjective inspection methods. Recognizing the need for efficient damage identification and the creation of 3D models for maintenance purposes, this paper introduces an innovative approach to the inspection of reinforced concrete bridges. The proposed methodology involves 3D reconstruction of a bridge, coupled with a concrete damage classification system based on severity. Notably, the analysis ensures objectivity through the implementation of deep learning for classifying concrete damage in UAV-captured images. A noteworthy aspect of this research is that, in the training models, a precision of over 90% is achieved for each type of concrete damage. This methodology serves as a valuable contribution to automating and streamlining concrete bridge inspections, aiming to reduce costs and enhance efficiency throughout its life cycle.

Keywords: UAV, Bridge, Concrete damage, SHM, CNN