Russian Scientists Develop System to Precisely Detect Manufacturing Defects
The technology helps industrial companies cut losses, prevent accidents, and improve equipment efficiency.
Researchers at Novosibirsk State Technical University have developed a domestically produced machine-vision system designed to monitor industrial production processes.
The system can be integrated into existing infrastructure without halting operations. It automatically detects defects on production lines with 94% accuracy, monitors compliance with safety requirements, and helps companies reduce operating costs.
Modular Monitoring Ecosystem
The modular ecosystem includes a device to improve network quality at production sites, a camera for defect detection and safety monitoring equipped with an AI module, a data-processing center that aggregates camera feeds, and a web-based control panel.
The system works as follows: the camera tracks every object on the line, while two neural networks analyze the video stream. Deviations from normal parameters are identified using heat maps, classifying defects into three categories: normal, dent, and crack.
Researchers are now working on further improving the platform to enable centralized management of the entire ecosystem.
