Deep Learning-Based Damage Detection in Concealed Cold-Formed Steel Structures using Ground Penetrating Radar

Cold-formed steel (CFS) is becoming increasingly prevalent in residential and commercial construction due to its cost effectiveness, high strength-to-weight ratio and fire resistance. As CFS buildings age and become more prevalent in hazard-prone areas, there's a need for a robust framework for structural condition assessment. Current assessment methods are limited by the presence of cladding on CFS structures.

To address this, Muhammad Taseer Ali of the University of Houston, proposes a two-part framework involving ground-penetrating radar (GPR) for data acquisition and deep learning, specifically the InternImage model, for data processing. The InternImage model utilizes a convolutional neural network (CNN) with deformable convolution DCNv3 to automate damage detection on concealed CFS structural members. The framework is demonstrated and evaluated for detecting various damage types on CFS load-bearing elements. The proposed approach has implications for automated assessment, particularly in post-disaster scenarios, providing direct access to concealed members without cladding removal. It also offers potential efficiency in time and cost for developing preventive maintenance strategies for aging CFS structures.

Muhammad Taseer Ali
University of Houston

Muhammad Taseer Ali has almost 10 years of experience in the construction industry. He has worked as a Structural Engineer in different companies in Pakistan, Middle East and the United States. He has managed engineering teams, executed structural design of cold-formed steel (CFS) buildings, and coordinated with cross-functional teams to deliver successful projects.

Currently, Taseer is a part of Structures and Artificial Intelligence Lab (SAIL), where he is pursuing his Ph.D. in Structural Engineering from the University of Houston under the supervision of Dr. Vedhus Hoskere. His research includes a deep learning-based framework for structural condition assessment and health monitoring of Cold-Formed Steel Structures.