Image classification on Post-Earthquake damage assessment: A case of the 2023 Kahramanmaraş earthquake

Abstract

Experts conduct damage assessments throughout the city in earthquake-prone areas to evaluate the destruction caused by the earthquake. Based on the ATC-20 Building Safety Values, the buildings impacted by the earthquake are categorized as “Inspected, Restricted Use, Unsafe”. Visual imagery captured both inside and outside the buildings is utilized to document the expedited identification of structural deficiencies and their underlying causes. Nevertheless, architects and engineers find the documentation, reporting, and decision-making process to be a time-consuming task. In the past ten years, extensive research has been carried out to reduce the duration of these procedures, specifically in the fields of construction and machine learning. This study investigates the application of machine learning in decision support systems, drawing on research on post-earthquake damage assessment. Post-earthquake damage assessment reports utilized CNN damage assessment algorithms to classify exterior images of buildings exhibiting “Inspected, Restricted Use, Unsafe” damage. The accuracy and loss values of various algorithms, including different AlexNet algorithms, the VGG19 algorithm, and the Resnet50 algorithm, were compared.

Keywords

• Machine Learning
• Post-Earthquake
• Image-Classification
• Damage Assessment
• Kahramanmaraş Earthquake