Automated Inspection for Rail Cars Using Computer Vision and Machine Learning

Discover how Canadian National Railways (CN) leverages computer vision and machine learning for automated rail car inspections in this 32-minute conference talk from the Toronto Machine Learning Series. Learn about CN's early adoption of AI technologies in railroad operations, focusing on their automated inspection system that uses strategically positioned cameras to capture images of rail cars from all angles. Explore the process of developing machine learning pipelines for defect detection, including challenges such as selecting representative training data, addressing image quality issues caused by varying weather and lighting conditions, and handling subjectivity in defect classification. Gain insights into innovative solutions like self-supervised learning techniques for identifying unique samples from unlabeled datasets. Examine a specific use case to understand how CN overcomes challenges to improve the performance of their automated inspection systems, demonstrating the potential of computer vision approaches in developing effective solutions for rail car inspection.