Cherepovets Steel Mill Comes Under the Watch of “SOVA”
Cherepovets Metallurgical Plant (CherMK), part of the Severstal group, has completed pilot tests of the SOVA digital video analytics system based on computer vision technology. The platform identifies workplace safety violations in real time.
Failure to wear a helmet or gloves, or a worker entering a hazardous area – these events are immediately transmitted to the monitoring system for rapid response.
Development and deployment of SOVA began in late 2024. The project was implemented internally by the plant’s own specialists, including the information technology and occupational safety departments.
Across several hundred cameras installed at the facility, engineers deployed ten computer vision models. The system successfully monitored workers’ presence in hazardous zones, supervised loading and unloading operations, tracked work at height and verified the use of personal protective equipment.
To ensure the security of the cameras themselves, the SOVA platform was integrated with the KOT software suite – also developed in Russia. The system allows engineers to configure customized threat-detection scenarios for each camera.
What Does This Mean for Industry?
The adoption of technologies such as SOVA reflects growing technological sovereignty in industrial software and signals a broader shift toward “smart manufacturing” in the metallurgical sector. The introduction of artificial intelligence into heavy industry is particularly significant because this sector has historically adopted innovations more slowly than other industries.
The technologies used in the SOVA project rely on a set of machine vision algorithms capable of conducting video monitoring and data analysis without direct human involvement. Video analytics automates four key security functions – detection, tracking, recognition and predictive analysis. One of the most advanced capabilities today is biometric-based facial recognition.
Industry Context
The example of CherMK is not unique within Russia. A number of mining and metals companies across the country are already applying machine vision systems to strengthen industrial safety.
One example is Norilsk Nickel. At the company’s facilities, computer vision systems monitor the use of personal protective equipment and track activity in hazardous zones. Cameras can detect workers without helmets or safety vests and automatically notify a dispatcher. As a result, safety violations dropped by 35% within the first months of operation. At Norilsk Nickel mines, the Antinaezd system is installed on self-propelled equipment. The platform identifies the presence of a person nearby and helps prevent collisions with pedestrians. The neural network analyzes thermal imaging signals, enabling operation in low-light conditions. At the Komsomolsky mine in the company’s Polar Division, a machine vision system also identifies oversized fragments of ore, reinforcement bars, wood and other foreign objects moving along conveyor belts. Artificial intelligence recognizes potentially dangerous objects listed in its database and alerts the operator to stop the conveyor and remove them.
Lebedinsky GOK, part of Metalloinvest, has deployed a video analytics system to monitor ore transportation through a continuous conveyor system. Cameras installed on unloading points, transfer stations, storage conveyors and main conveyor lines identify defects in conveyor belts in real time, including punctures, cuts and misalignment. Two additional cameras detect oversized ore fragments. As a result, downtime on conveyors and crushers has been reduced by nearly half.
At the Udachny underground mine and the Yubileyny open pit operated by ALROSA, engineers have tested a truck loading monitoring system. A lidar sensor creates a three-dimensional model of a haul truck’s body, while a neural network calculates the volume of rock loaded and the degree of fill. From the control center, dispatchers can adjust equipment operators’ actions in real time. According to test results, the productivity of mining haul trucks increased by 10%, equivalent to nearly 12,000 tonnes of additional diamond-bearing ore transported.
In the future, Severstal plans to scale the SOVA system across other enterprises within its industrial group and integrate video analytics with industrial internet of things platforms. According to experts, within the next five to ten years computer vision and video analytics systems could become a standard element of digital production ecosystems.
