A Deep Learning Application for Detecting Facade Tile Degradation

  • Po-Hsiang Shih
  • Kuang-Hui ChiEmail author
Conference paper
Part of the Advances in Intelligent Systems and Computing book series (AISC, volume 1026)


Facade tiles of buildings are likely to weaken, crack, or fall off due to aging or out of natural causes such as temperature variations during daytime and nighttime and earthquakes. Tile spalling of tall buildings often leads to accidents or even severe casualties. In view that a routine thorough inspection is costly, this study aims to develop a cost-effective means to detect facade tile degradation of tall buildings through machine learning. We leverage a drone to film outer walls of high-rise buildings at several dozens of sites, from which training data are produced for learning and validation. We resort to a convolutional neural network with deep learning capabilities that is trained with sufficient knowledge to identify hazardous conditions of cracked tiles in two or three levels. Core to our implementation is Jetson TX2—an embedded system—which is programmed in light of AlexNet over Keras and TensorFlow, open-source libraries for deep neural network programming. To heighten learning quality subject to limited amount of training data, image preprocessing involving gray-level transformation, thresholding, and morphological operations is introduced. Experimental results corroborate that our scheme achieves a correct classification rate of over 86%. Our development serves a moderate approach to deep learning in daily contexts, a practical scenario over which to inspire other applications.


Machine learning Deep learning Convolutional neural network Defect detection Image processing AlexNet 



This work was supported by the Ministry of Science and Technology, ROC, under grant MOST 107-2221-E-224-051.


  1. 1.
    Krizhevsky, A., Sutskever, I., Hinton, G.E.: ImageNet classification with deep convolutional neural networks. In: Proceedings 25th International Conference Neural Information Processing Systems, vol. 1, pp. 1097–1105 (2012)Google Scholar
  2. 2.
    Lin, M., Chen, Q., Yan, S.: Network in network. In: Proceedings International Conference Learning Representations (2013)Google Scholar
  3. 3.
    Schmidhuber, J.: Deep learning in neural networks: an overview. Neural Netw. 61, 86–117 (2015)CrossRefGoogle Scholar
  4. 4.
    Schroff, F., Kalenichenko, D., Philbin, J.: FaceNet: a unified embedding for face recognition and clustering. In: Proceedings 2015 IEEE Conference Computer Vision and Pattern Recognition (2015)Google Scholar
  5. 5.
    Vu, H.T., Huang, C.-C.: A multi-task convolutional neural network with spatial transform for parking space detection. In: Proceedings 2017 IEEE International Conference Image Processing (2017)Google Scholar
  6. 6.
    Jaderberg, M, Simonyan, K., Zisserman, A., Kavukcuoglu, K.: Spatial transformer networks. arXiv: 1506.02025, (2016)

Copyright information

© Springer Nature Switzerland AG 2020

Authors and Affiliations

  1. 1.Department of Electrical EngineeringNational Yunlin University of Science and TechnologyDouliuTaiwan

Personalised recommendations