Rail Track Defects Detection Using Convolution Neural Network

Rail track defects detection is a critical aspect of ensuring the safety and efficiency of railway systems. This abstract introduces a methodology for detecting such defects using Convolutional Neural Networks (CNNs). The process begins with preprocessing input images to enhance their quality and relevance for analysis. Subsequently, a CNN architecture is employed, comprising multiple layers designed for feature extraction and classification. Various training options are explored to optimize the CNN's performance, including dataset preprocessing techniques. The deep learning algorithm is then applied to train the CNN model on labelled data, enabling it to distinguish between healthy and unhealthy rail tracks. If the CNN classifies a track as unhealthy, further classification is performed to identify specific defects. The accuracy of the detection system is evaluated to assess its reliability in real-world scenarios. By leveraging CNNs, this approach offers a promising solution for automating rail track inspection processes, potentially improving safety and reducing maintenance costs in railway infrastructure management.

Keywords: Rail Track Images Dataset, Pre-Processing, Deep Learning Algorithm, Convolution Neural Network, Defects Identification Accuracy.