University of Waterloo

Rail Segmentation and Defect Detection

Xuefeng Ni, Visiting graduate student

Dec 3, 2022, 11:30am EC4-2101A

Millions of kilometers of rails need to be inspected throughout the world. Rail surface damage inspection based on machine vision is an efficient means of railway maintenance because of the rapidity and consistency of image acquisition, and our particular research goal is to realize an all-weather on-board damage inspection for the rails with high reliability. Our research focus is to design damage detection algorithms based on image processing and deep learning. In this seminar, some practical design challenges will be discussed, followed by methods and directions for rail surface analysis and damage inspection.