Abstract
Crack detection on road or building surface is normally inspected manually by specialist. It consumes a lot of time and the inspection result might be different depending on the specialist experience and knowledge. In this paper, an automated crack segmentation model built using DeepLab model is proposed where transfer learning is being utilized. The model is trained on the dataset from DeepCrack which consists of 300 training images and 237 testing images. 3 models are trained with different value of training step and training rate. The models are then evaluated using the mean intersection-over-union metrics and managed to achieve value around 0.75 for mean intersection-over-union. 10 images also chosen and the precision and recall value for each of the images are calculated and plotted on a graph. The segmentation result of the DeepLab model was used to compare with the segmentation result of Otsu’s method in detecting cracks.
Keywords: crack segmentation, DeepLab, transfer learning
Authors
Voon Zhen Cheng [1] ; Chaw Jun Kit [2]
[1][2] Department of Computer Science and Embedded Systems, Faculty of Computing and Information Technology, Tunku Abdul Rahman University of Management and Technology, Kuala Lumpur, Malaysia
[1] vzhencheng97@gmail.com ; [2] chawjk@tarc.edu.my
Cite Me
Plain Text:
Z.C.Voon, J.K.Chaw, "Crack Segmentation Using Deeplab," International Conference on Digital Transformation and Applications (ICDXA) 2020, 2020, pp. 106-111, doi: https://doi.org/10.56453/icdxa.2020.1011.
BibTex:
@INPROCEEDINGS{ICDXA2020T204,
author={Voon, Zhen Cheng and Chaw, Jun Kit},
booktitle={International Conference on Digital Transformation and Applications (ICDXA) 2020},
title={Crack Segmentation Using Deeplab},
year={2020},
volume={},
number={},
pages={106-111},
doi={https://doi.org/10.56453/icdxa.2020.1011}}
