Deep Learning for Physical Damage Detection in Buildings: A Comparison of Transfer Learning Methods

Year 2023, Volume: 18 Issue: 2, 291 - 299, 01.09.2023

Betül Bektaş Ekici , Saltuk Taha Ustaoğlu

https://doi.org/10.55525/tjst.1291814

Abstract

The detection of physical damage in buildings is a critical task in ensuring the safety and integrity of structures. In this study, the effectiveness of deep learning methods for detecting physical damage in buildings, specifically focusing on cracks, defects, moisture, and undamaged classes was investigated. Transfer learning methods, including VGG16, GoogLeNet, and ResNet50, were used to classify a dataset of 7200 images. The dataset was split into training, validation, and testing sets, and the performance of the models was evaluated by using metrics such as accuracy, precision, recall, and F1-score. Results show that all three models achieved high accuracy on the test set, with VGG16 and ResNet50 outperforming GoogLeNet. Additionally, precision, recall, and F1-score metrics indicate strong performance across all classes, with VGG16 and ResNet50 achieving particularly high scores. It is demonstrated the effectiveness of deep learning methods for physical damage detection in buildings and provides insights into the comparative performance of transfer learning methods.

Keywords

Structural damage classification, deep learning, convolutional neural networks, transfer learning.

Supporting Institution

---

Project Number

---

Thanks

---

Binalarda Fiziksel Hasar Tespiti için Derin Öğrenme: Transfer Öğrenme Yöntemlerinin Karşılaştırılması

Abstract

Binalardaki fiziksel hasarın tespiti, yapıların güvenliğini ve bütünlüğünü sağlamada kritik bir görevdir. Bu çalışmada, özellikle çatlaklar, kusurlar, nem ve hasarsız sınıflara odaklanarak binalardaki fiziksel hasarı tespit etmek için derin öğrenme yöntemlerinin etkinliği araştırılmıştır. VGG16, GoogLeNet ve ResNet50 dahil olmak üzere transfer öğrenme yöntemleri, 7200 görüntüden oluşan bir veri kümesini sınıflandırmak için kullanılmıştır. Veri kümesi eğitim, doğrulama ve test kümelerine ayrılmış ve modellerin performansı doğruluk, kesinlik, geri çağırma ve F1-skoru gibi ölçütler kullanılarak değerlendirilmiştir. Sonuçlar, üç modelin de test setinde yüksek doğruluk elde ettiğini, VGG16 ve ResNet50'nin GoogLeNet'ten daha iyi performans gösterdiğini ortaya koymuştur. Ayrıca, hassasiyet, geri çağırma ve F1-skoru ölçümleri tüm sınıflarda güçlü performans gösterirken, VGG16 ve ResNet50 özellikle yüksek puanlar elde etmiştir. Binalarda fiziksel hasar tespiti için derin öğrenme yöntemlerinin etkinliği gösterilmiş ve transfer öğrenme yöntemlerinin karşılaştırmalı performansına ilişkin içgörüler sağlanmıştır.

Keywords

Yapısal hasar sınıflandırması, derin öğrenme, evrişimli sinir ağları, transfer öğrenme.

Project Number

---

References

• Ekici BB. Detecting damaged buildings from satellite imagery. J. Appl. Remote Sens 2021; 15(3): 032004.
• Sharma N, Sharma R, Jindal N. Machine learning and deep learning applications-a vision. Glob. Transitions Proc 2021; 2(1): 24–28.
• Iman M, Arabnia HR, Rasheed K. A review of deep transfer learning and recent advancements. Technol 2023; 11(2): 40.
• Zhuang F, Qi Z, Duan K, Xi D, Zhu Y, Zhu H, Xiong H, He Q. A comprehensive survey on transfer learning. Proc. IEEE 2021; 109(1): 43–76.
• “Visual Geometry Group - University of Oxford.” https://www.robots.ox.ac.uk/~vgg/ (accessed May 02, 2023).
• “ILSVRC2014 Results.” https://image-net.org/challenges/LSVRC/2014/results (accessed May 02, 2023).
• “Deep Residual Networks (ResNet, ResNet50) 2023 Guide - viso.ai.” https://viso.ai/deep-learning/resnet-residual-neural-network/ (accessed May 02, 2023).
• Kung RY, Pan NH,Wang CCN, Lee PC. Application of deep learning and unmanned aerial vehicle on building maintenance. Adv. Civ. Eng 2021; 2021: 5598690.
• Flah M, Suleiman AR, Nehdi ML. Classification and quantification of cracks in concrete structures using deep learning image-based techniques. Cem Concr Compos 2020; 114: 103781.
• Wang N, Zhao X, Zou Z, Zhao P, Qi F. Autonomous damage segmentation and measurement of glazed tiles in historic buildings via deep learning. Comput Civ Infrastruct Eng 2020; 35(3): 277–291.
• Yang J, Zhang L, Chan C, Li Y, Li R, Wang G, Jiang S, Zeng Z. A hierarchical deep convolutional neural network and gated recurrent unit framework for structural damage detection. Inf Sci 2020; 540: 117–130.
• Wang N, Zhao X, Wang L, Zou Z. Novel system for rapid investigation and damage detection in cultural heritage conservation based on deep learning. J Infrastruct Syst 2019; 25(3):04019020.
• Nex F, Duarte D, Tonolo FG, Kerle N. Structural building damage detection with deep learning: assessment of a state-of-the-art cnn in operational conditions: Remote Sens 2019; 11(23): 2765.
• Jiang Y, Pang D, Li C, Yu Y, Cao Y. Two-step deep learning approach for pavement crack damage detection and segmentation. Int. J. Pavement Eng 2022:1-14. doi: 10.1080/10298436.2022.2065488.
• Lin YZ, Nie ZH, Ma HW. Structural damage detection with automatic feature-extraction through deep learning. Comput Civ Infrastruct Eng 2017; 32(12): 1025–1046.
• Dais D, Bal İE, Smyrou E, Sarhosis V. Automatic crack classification and segmentation on masonry surfaces using convolutional neural networks and transfer learning. Autom Constr; 125: 103606.
• Perez H, Tah JHM, Mosavi A. Deep learning for detecting building defects using convolutional neural networks. Sensors 2019; 19 (16): 3556.
• Teng S, Chen X, Chen G, Cheng L. Structural damage detection based on transfer learning strategy using digital twins of bridges. Mech Syst Signal Process 2023; 191:110160.
• Elghaish F, Talebi S, Abdellatef E, Matarneh ST, Reza Hosseini M, Wu S, Mayouf M, Hajirasouli A, Nguyen TQ. Developing a new deep learning cnn model to detect and classify highway cracks. J Eng Des Technol 2022; 20 (4): 993–1014.
• Feng C, Zhang H, Wang S, Li Y, Haoran W, Yan F. Structural Damage Detection using deep convolutional neural network and transfer Learning 2019: 23(1); 1-10.
• Gulgec, NS, Takáč M, Pakzad, SN. Structural Damage Detection Using Convolutional Neural Networks. In: 35th IMAC, A conference and exposition on structural dynamics; Ja 30-Feb 2 2016; CA, USA: Cham, Germany: Springer. pp. 331-337.
• Eltouny KA, Liang X. Bayesian-optimized unsupervised learning approach for structural damage detection. Comput Civ Infrastruct Eng 2021; 36 (10): 1249–1269.
• S. T. Ustaoğlu, “Detection of Building Physics Problems With Convolutional Neural Networks. MSc, Fırat University, Elazığ, Türkiye, 2023.
• “Transfer Learning Using Pretrained Network - MATLAB & Simulink.” https://www.mathworks.com/help /deeplearning/ug/transfer-learning-using-pretrained-network.html (accessed May 02, 2023).