AI-powered Machine Vision-Based Inspection Systems (MVIS)
Kartavya Desk Staff
Source: News on Air
Context: Indian Railways and DFCCIL have signed an MoU to deploy AI-powered Machine Vision-Based Inspection Systems (MVIS) for real-time detection of rolling stock defects.
About AI-powered Machine Vision-Based Inspection Systems (MVIS):
• What is MVIS? MVIS is an AI–ML integrated visual inspection platform developed to monitor and detect defects in freight train rolling stock using high-resolution cameras and computer vision. It automates inspection, enhances maintenance efficiency, and reduces the risk of accidents.
• MVIS is an AI–ML integrated visual inspection platform developed to monitor and detect defects in freight train rolling stock using high-resolution cameras and computer vision.
• It automates inspection, enhances maintenance efficiency, and reduces the risk of accidents.
• Developed by: DFCCIL in collaboration with IISc Bengaluru and start-up L2M.
• Objectives of MVIS:
• Detect abnormal hanging parts, broken springs, missing bolts, EM pad damage, hot axles, etc. Provide real-time alerts to prevent accidents and service delays. Replace manual inspection with an automated, accurate, and fatigue-free system. Enable preventive maintenance to reduce cascading failures in rail operations.
• Detect abnormal hanging parts, broken springs, missing bolts, EM pad damage, hot axles, etc.
• Provide real-time alerts to prevent accidents and service delays.
• Replace manual inspection with an automated, accurate, and fatigue-free system.
• Enable preventive maintenance to reduce cascading failures in rail operations.
• How MVIS Works?
• High-Speed Cameras (Area & Line Scan): Mounted on trackside to capture images of moving trains at speeds up to 100 km/h. AI/ML-Powered Analysis: Uses YOLOv8 & CNN models to detect and classify components as defective or non-defective. OCR integration for wagon number detection. Data Processing Units (DPUs): Process real-time footage; synchronized with NTP servers to avoid timestamp mismatches. GUI Dashboard: Cloud-based portal for defect reports, train-wise metrics, and maintenance action logs
• High-Speed Cameras (Area & Line Scan): Mounted on trackside to capture images of moving trains at speeds up to 100 km/h.
• AI/ML-Powered Analysis: Uses YOLOv8 & CNN models to detect and classify components as defective or non-defective. OCR integration for wagon number detection.
• Uses YOLOv8 & CNN models to detect and classify components as defective or non-defective.
• OCR integration for wagon number detection.
• Data Processing Units (DPUs): Process real-time footage; synchronized with NTP servers to avoid timestamp mismatches.
• GUI Dashboard: Cloud-based portal for defect reports, train-wise metrics, and maintenance action logs
• Key Features of MVIS:
• Multi-camera configuration (upper, lower, undercarriage) to capture all critical components. Monochrome cameras used for faster processing and better clarity in defect detection. LED Lighting ensures proper imaging in day-night and poor visibility conditions. Real-time alerting system enables immediate remedial action by maintenance teams. Scalable architecture to handle large data, train formations, and national deployment.
• Multi-camera configuration (upper, lower, undercarriage) to capture all critical components.
• Monochrome cameras used for faster processing and better clarity in defect detection.
• LED Lighting ensures proper imaging in day-night and poor visibility conditions.
• Real-time alerting system enables immediate remedial action by maintenance teams.
• Scalable architecture to handle large data, train formations, and national deployment.