A Deep Learning Approach for Robust Automatic Crack Detection using unsupervised multi-scale CNN

Abstract

In this paper a deep learning based approach has been used for automatic crack detection. Generally, railway track crack detection is performed by using   ultra-supersonic expertise, which involves manual crack detection of bulk quantity of data. This technique meets deficiencies of minimal competence, a very large detection life cycle, and this technique needs a skilled specialized with high-level of hands-on experiences on railway track crack data analysis. In this research, we develop a Robust Automatic Railway Track Crack Detection using unsupervised multi-scale Convolutional Neural Network based Deep Learning.