An MOU has been signed by Indian Railways and DFCCIL to deploy AI-powered Machine Vision-Based Inspection Systems (MVIS)

What does MVIS stand for?

MVIS is an integrated visual inspection platform utilizing AI–ML to monitor and identify defects in freight train rolling stock through high-resolution cameras and computer vision.

It streamlines inspection, improves maintenance effectiveness, and minimizes accident risk.

Created by: DFCCIL alongside IISc Bengaluru and start-up L2M.

Goals of MVIS:

Identify irregular hanging components, damaged springs, absent bolts, EM pad deterioration, overheated axles, etc.

Deliver instant notifications to avert accidents and avoid service interruptions.

Substitute manual evaluation with a precise, automated, and tireless system.

Implement preventive maintenance to minimize cascading failures in rail operations.

How does MVIS operate?

High-Speed Cameras (Area & Line Scan): Positioned along the track to photograph trains in motion at velocities reaching 100 km/h.

An analysis driven by AI/ML:

Employs YOLOv8 and CNN models to identify and categorize components as either defective or non-defective.

Integration of OCR for detecting wagon numbers.

Data Processing Units (DPUs): Handle live video streams; aligned with NTP servers to prevent timestamp discrepancies.

GUI Interface: Online platform for issue reports, train-specific data, and upkeep activity records

Principle Characteristics of MVIS:

Multi-camera setup (top, bottom, chassis) to record all essential elements.

Monochrome cameras employed for quicker processing and enhanced clarity in identifying defects.

LED Lighting guarantees effective imaging in both daytime and nighttime as well as in low visibility situations.

A real-time notification system allows maintenance teams to take prompt corrective measures.

Architecture that can scale to manage extensive data, train models, and support nationwide implementation.