Related Experiment Video
Updated: Jul 4, 2025

Kinematic History of a Salient-recess Junction Explored through a Combined Approach of Field Data and Analog Sandbox Modeling
Published on: August 5, 2016
Railway track surface faults dataset
Asfar Arain1, Sanaullah Mehran1, Muhammad Zakir Shaikh1,2
1NCRA MUET, NCRA Condition Monitoring Systems Lab, Mehran University of Engineering and Technology, Jamshoro, Sindh, Pakistan.
A new dataset of railway track surface defects, captured by EKENH9R cameras, aids research in railway maintenance and computer vision. This resource supports developing Machine Learning (ML) and Deep Learning (DL) for automated track inspection.
Area of Science:
- Civil Engineering
- Computer Vision
- Data Science
Background:
- Railway infrastructure maintenance is vital for transportation safety and efficiency.
- Track surface defects like cracks and spallings challenge track integrity.
- Existing research requires comprehensive datasets for advanced analysis.
Purpose of the Study:
- Introduce a novel dataset of railway track surface faults.
- Provide a valuable resource for railway maintenance and computer vision research.
- Facilitate the development of ML/DL algorithms for defect detection.
Main Methods:
- Collected data using EKENH9R cameras on a railway inspection vehicle.
- Ensured diverse real-world fault representation under various conditions.
- Provided detailed annotations and metadata for precise classification.
Main Results:
- A comprehensive dataset of diverse railway track surface faults is now available.
- The dataset includes images under varied environmental and lighting conditions.
- Annotations enable precise fault classification and severity assessment.
Conclusions:
- The dataset is a significant asset for ML/DL and image processing in railway maintenance.
- It supports the development of automated inspection and predictive maintenance systems.
- Encourages community utilization for advancing track condition monitoring research.
More Related Videos
Related Concept Videos
Adjusting a Traverse
Errors and Mistakes in Surveying
Distance Corrections
Errors in Taping
Profile Leveling and Cross Sections
Latitudes and Departures

