Railway Track Defect Detection using Machine Vision

Abstract

Defect Detection on Rail head surface is a major issue for railway transportation. Rail maintenance of railway track is necessary to avoid any railway accidents due to the defects on railway track surface. Since rail surface defects have such a wide range of characteristics, it’s difficult to categorize them using machine learning. In this project machine vision-based defect detection was implemented using Deep learning-based object detection and semantic segmentation. Image enhancement techniques were also implemented which helped in making an efficient deep learning model in training.