OMK has implemented AI-based service for scrap metal quality control

The United Metallurgical Company (OMK) has introduced a digital service at its Vyksa Metallurgical Works, powered by machine vision and neural networks, to automatically determine the type, quality, and purity of steel scrap delivered to the facility. The system is operational in integrated casting and rolling facility, where scrap metal is processed for steel smelting.

Industrial cameras installed in the workshops continuously photograph and record incoming raw materials, transmitting data to a system that analyzes every unloaded layer of scrap in railcars or trucks.

Three computer vision models have been integrated: The first model analyzes the video stream and identifies layers of unloaded scrap. The second evaluates scrap layers in real time for contamination and sends automated alerts to stop unloading if necessary. The third verifies compliance of the raw materials with GOST standards specified in documentation.

The service is also being trained to recognize and block the unloading of prohibited items, such as potentially explosive objects e.g., gas cylinders, barrels, and others. This functionality will soon be incorporated into the production process.

The system generates a report for each unloaded vehicle, including layer-by-layer photos of the scrap. These reports are stored and can be shared with suppliers to validate certification.

The solution aims to optimize and accelerate raw material unloading, save time for employees involved in acceptance procedures, resolve disputes with suppliers more efficiently, and shorten product acceptance timelines.

Ilya Dzyub, OMK’s Chief Architect for Digital Technology Development, commented:

“Optimization is the cornerstone of all automation processes in production. Computer vision has eliminated the need for Quality engineers to work near elevating mechanisms and streamlined quality control without relying on scarce specialists. We delegate routine tasks to machines, freeing up employees for critical roles. However, machines won’t fully replace humans—operators still monitor the system’s performance. This synergy reduces costs and enhances raw material quality.”