Railway Track Surface Faults Dataset

This dataset is related to railway track faults which includes different types of railway track surface faults. The data was acquired as a video through two EKEN-H9R cameras mounted on both sides of the railway inspection vehicle. The cameras were capable of recording videos at 120 FPS and with a FOV (field of view) of approximately 14 inches. Unnecessary parts of the collected videos were removed in the second step. Then, the necessary frames were extracted from the video. After extracting required frames out of videos manual labeling of the images was done. The established datasets includes classes: Grooves, Joints, Cracks, Flakings, Shellings, Spallings, and Squats.

本数据集聚焦铁路轨道表面故障，涵盖多种类型的轨道表面缺陷。数据采集环节，通过安装于轨道巡检车两侧的两台EKEN-H9R相机以视频形式完成数据采集，该相机支持120帧每秒（FPS）的视频录制，视场角（FOV, Field of View）约为14英寸。后续先对采集所得视频进行预处理，剔除冗余片段；随后从处理后的视频中提取所需帧图像；最终完成提取图像的人工标注工作。本数据集涵盖的故障类别如下：凹槽（Grooves）、轨缝（Joints）、裂纹（Cracks）、片状剥落（Flakings）、壳状剥落（Shellings）、碎裂掉块（Spallings）以及踏面凹坑（Squats）。

搜集汇总

背景与挑战

背景概述

该数据集包含铁路轨道表面多种故障类型的图像，如裂纹、剥落等，通过高帧率摄像机采集并经过手动标注处理，适用于机械工程和系统故障诊断等领域的研究。