Using Data Augmentation for automated recognition of façade anomalies
DOI:
https://doi.org/10.46421/entac.v20i1.5843Keywords:
Automated inspection., Visual assets., Unbalanced base., Machine learningAbstract
In recent years, there has been an increase in the use of Artificial Intelligence (AI) for automated image analysis. However, there are some limitations regarding small data sets. Aiming to minimize this limitation, this study evaluates using Data Augmentation (DA) techniques to create new automated image recognition models. The research strategy used was an experimental simulation based on (i) refinement of the database of images of concrete facades, (ii) development of a DA code to expand the database, (iii) training and testing images using web platforms with pre-trained networks; and (iv) analysis of results through performance indicators. The results indicated that the “Model with contrast method,” using ResNet and AlexNet algorithms, achieved 67.3% precision and 94.6% Recall.
References
SILVA, A.S.; COSTA, D. B. Análise do uso de tecnologias digitais para identificação automatizada de patologias em construções.
ENCONTRO NACIONAL DE TECNOLOGIA DO AMBIENTE CONSTRUÍDO, v. 19, p. 1-14, 2022.
STOCHINO, F.; FADDA, M. L.; MISTRETTA, F. Low cost condition assessment method for existing RC bridges. Engineering Failure
Analysis, v. 86, p. 56-71, 2018.
BARBOSA, R. L. de B.. Sistema de Deep Learning para detecção de trincas em concreto. [s.l.] Universidade de Pernambuco, 2023.
ALI, S. B. et al. Wall crack detection using transfer learning-based CNN models. In: 2020 IEEE 17th India Council International
Conference (INDICON). IEEE, 2020. p. 1-7.
ISLAM, M. M. et al. CNN based on transfer learning models using data augmentation and transformation for detection of concrete
crack. Algorithms, v. 15, n. 8, p. 287, 2022.
KIM, B. et al. Surface crack detection using deep learning with shallow CNN architecture for enhanced computation. Neural
Computing and Applications, v. 33, p. 9289-9305, 2021.
KAUFMAN, D. A inteligência artificial irá suplantar a inteligência humana?. ESTAÇÃO DAS LETRAS E CORES EDI, 2019.
SHORTEN, C.; KHOSHGOFTAAR, T. M.; FURHT, B. Text data augmentation for deep learning. Journal of big Data, v. 8, p. 1-34, 2021.
SILVA, A. S.; MELO, Roseneia Rodrigues Santos; COSTA, Dayana Bastos. Automated facade inspection: Application and challenge in
using Artificial Intelligence for construction defect recognition. In: XX International Conference on Building Pathology and
Constructions Repair (CINPAR), p. XXXX, 2024.
OTTONI, A. L. C. et al. Métodos para recomendação de hiperparâmetros de aprendizado de máquina na classificação de imagens
da construção civil. 2022.
OTTONI, A. L. C.; NOVO, M. S. A deep learning approach to vegetation images recognition in buildings: a hyperparameter tuning
case study. IEEE Latin America Transactions, v. 19, n. 12, p. 2062-2070, 2021.
GUO, J. et al. Façade defects classification from imbalanced dataset using meta learning‐based convolutional neural network.
Computer‐Aided Civil and Infrastructure Engineering, v. 35, n. 12, p. 1403-1418, 2020.
CHENG, J.C. AND WANG, M. Automated detection of sewer pipe defects in closed-circuit television images using deep learning
techniques. Automation in Construction, 95, p.155-171, 2018.
LEE, K., LEE, S., & KIM, H. Y. (2022). Bounding-box object augmentation with random transformations for automated defect
detection in residential building façades. Automation in Construction, 135, 104138. https://doi.org/10.1016/j.autcon.2022.104138
MAHARANA, K.; MONDAL, Surajit; NEMADE, Bhushankumar. A review: Data pre-processing and data augmentation techniques.
Global Transitions Proceedings, v. 3, n. 1, p. 91-99, 2022.
OTTONI, A.L.C., DE AMORIM, R.M., NOVO, M.S. et al. Tuning of data augmentation hyperparameters in deep learning to building
construction image classification with small datasets. Int. J. Mach. Learn. & Cyber. 14, 171–186 (2022). https://doi.org/10.1007/
s13042-022-01555-1.
KIM, D.; CHOE, S.; ZHANG, S. Recognition of adherent polychaetes on oysters and scallops using Microsoft Azure Custom Vision.
Electronic Research Archive, v. 31, n. 3, p. 1691-1709, 2023.
LIAKHOVICH, Olga; MBEMBA, Claudius. Food classification with custom vision service. 2017.
DANIEL, E. R. Wildfire smoke detection with computer vision. arXiv preprint arXiv:2301.05070, 2023.
SHANDILYA, Shishir Kumar et al. YOLO-based segmented dataset for drone vs. bird detection for deep and machine learning
algorithms. Data in Brief, v. 50, p. 109355, 2023.
SILVA, A. S. GONZAGA, L.G., MELO, R.R.S, COSTA, D.B. Modelo de aprendizado de máquina para inspeção automatizada de
fachadas de paredes de concreto. SIMPÓSIO BRASILEIRO DE GESTÃO E ECONOMIA DA CONSTRUÇÃO, v. 13, p. 1-9, 2023.
TAYLOR, L.; NITSCHKE, G. Improving deep learning with generic data augmentation. In: 2018 IEEE symposium series on
computational intelligence (SSCI). IEEE, 2018. p. 1542-1547.
MEDEIROS, L. C. Aplicação de Aprendizado Profundo Na Classificação de Imagens de Patologias Da Construção Civil: Análise de
Aumento de Dados Para Bancos de Dados Desbalanceados. 2023. Tese (Bacharelado em Ciências exatas e tecnológicas) -
Universidade Federal do Recôncavo da Bahia, Cruz das Almas, 2023.
