Categorization of Post-Earthquake Damages in RC Structural Elements with Deep Learning Approach

Mertcan Yilmaz; Gamze Dogan; Musa Hakan Arslan; Alper Ilki

doi:10.1080/13632469.2024.2302033

Categorization of Post-Earthquake Damages in RC Structural Elements with Deep Learning Approach

Mertcan Yilmaz, Gamze Dogan^*, Musa Hakan Arslan, Alper Ilki

^*Corresponding author for this work

Department of Civil Engineering

Research output: Contribution to journal › Article › peer-review

6 Citations (Scopus)

Abstract

The aim of this study was to develop an innovative deep learning based intelligent software (DamageNet) and its mobile applications to classify seismic damage of Reinforced Concrete (RC) elements. Images of 2455 damaged elements that have been exposed to different destructive earthquakes were collected from the “datacenterhub” database. The DamageNet algorithm has been compared with the pretrained convolutional neural networks (CNN) algorithms (VGG16, ResNet-50, MobileNetV2 and EfficientNet) according to performance metrics. With the other models, a maximum test success of 89% was achieved, while with DamageNet a test success of 92% was achieved in damage classification. The mobile application developed based on the DamageNet model was tested in the field after the earthquakes (M_w:7.7 and M_w:7.6) in Kahramanmaraş/Turkey and classification success of 88% was obtained.

Original language	English
Pages (from-to)	2620-2651
Number of pages	32
Journal	Journal of Earthquake Engineering
Volume	28
Issue number	9
DOIs	https://doi.org/10.1080/13632469.2024.2302033
Publication status	Published - 2024

Bibliographical note

Publisher Copyright:
© 2024 Taylor & Francis Group, LLC.

Keywords

Damage
DamageNet
convolutional neural network
damage assessment
earthquake

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1080/13632469.2024.2302033

Cite this

@article{27a8e0d4c36b4401bba7b2d528401b67,

title = "Categorization of Post-Earthquake Damages in RC Structural Elements with Deep Learning Approach",

abstract = "The aim of this study was to develop an innovative deep learning based intelligent software (DamageNet) and its mobile applications to classify seismic damage of Reinforced Concrete (RC) elements. Images of 2455 damaged elements that have been exposed to different destructive earthquakes were collected from the “datacenterhub” database. The DamageNet algorithm has been compared with the pretrained convolutional neural networks (CNN) algorithms (VGG16, ResNet-50, MobileNetV2 and EfficientNet) according to performance metrics. With the other models, a maximum test success of 89% was achieved, while with DamageNet a test success of 92% was achieved in damage classification. The mobile application developed based on the DamageNet model was tested in the field after the earthquakes (Mw:7.7 and Mw:7.6) in Kahramanmara{\c s}/Turkey and classification success of 88% was obtained.",

keywords = "Damage, DamageNet, convolutional neural network, damage assessment, earthquake",

author = "Mertcan Yilmaz and Gamze Dogan and Arslan, {Musa Hakan} and Alper Ilki",

note = "Publisher Copyright: {\textcopyright} 2024 Taylor & Francis Group, LLC.",

year = "2024",

doi = "10.1080/13632469.2024.2302033",

language = "English",

volume = "28",

pages = "2620--2651",

journal = "Journal of Earthquake Engineering",

issn = "1363-2469",

publisher = "Taylor and Francis Ltd.",

number = "9",

}

TY - JOUR

T1 - Categorization of Post-Earthquake Damages in RC Structural Elements with Deep Learning Approach

AU - Yilmaz, Mertcan

AU - Dogan, Gamze

AU - Arslan, Musa Hakan

AU - Ilki, Alper

PY - 2024

Y1 - 2024

N2 - The aim of this study was to develop an innovative deep learning based intelligent software (DamageNet) and its mobile applications to classify seismic damage of Reinforced Concrete (RC) elements. Images of 2455 damaged elements that have been exposed to different destructive earthquakes were collected from the “datacenterhub” database. The DamageNet algorithm has been compared with the pretrained convolutional neural networks (CNN) algorithms (VGG16, ResNet-50, MobileNetV2 and EfficientNet) according to performance metrics. With the other models, a maximum test success of 89% was achieved, while with DamageNet a test success of 92% was achieved in damage classification. The mobile application developed based on the DamageNet model was tested in the field after the earthquakes (Mw:7.7 and Mw:7.6) in Kahramanmaraş/Turkey and classification success of 88% was obtained.

AB - The aim of this study was to develop an innovative deep learning based intelligent software (DamageNet) and its mobile applications to classify seismic damage of Reinforced Concrete (RC) elements. Images of 2455 damaged elements that have been exposed to different destructive earthquakes were collected from the “datacenterhub” database. The DamageNet algorithm has been compared with the pretrained convolutional neural networks (CNN) algorithms (VGG16, ResNet-50, MobileNetV2 and EfficientNet) according to performance metrics. With the other models, a maximum test success of 89% was achieved, while with DamageNet a test success of 92% was achieved in damage classification. The mobile application developed based on the DamageNet model was tested in the field after the earthquakes (Mw:7.7 and Mw:7.6) in Kahramanmaraş/Turkey and classification success of 88% was obtained.

KW - Damage

KW - DamageNet

KW - convolutional neural network

KW - damage assessment

KW - earthquake

UR - http://www.scopus.com/inward/record.url?scp=85182709000&partnerID=8YFLogxK

U2 - 10.1080/13632469.2024.2302033

DO - 10.1080/13632469.2024.2302033

M3 - Article

AN - SCOPUS:85182709000

SN - 1363-2469

VL - 28

SP - 2620

EP - 2651

JO - Journal of Earthquake Engineering

JF - Journal of Earthquake Engineering

IS - 9

ER -

Categorization of Post-Earthquake Damages in RC Structural Elements with Deep Learning Approach

Abstract

Bibliographical note

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this