Wooden Sleeper Deterioration Evaluation System Using Image Analysis of Video

Abstract

This paper describes the development of a low-cost, simple system for inspecting wooden sleepers using images taken with a forward-facing handheld camera placed at the front of a train. This system is expected to save labor for inspecting track facilities. The system uses deep learning to classify the deterioration of wooden sleepers from images. In this paper, we report the outline of the system: classification accuracy, and verification results to see if the progress of deterioration can be assessed.