Neural Networks to Monitor Roof Safety in St. Petersburg This Winter
The city will deploy AI-powered systems to automatically detect hazardous rooftops, aiming to prevent falling snow and icicles before they pose a threat.
The State Administrative and Technical Inspectorate of St. Petersburg will roll out an artificial intelligence monitoring system during the 2025–2026 winter season. The technology will track how promptly owners of non-residential buildings clear snow and icicles from rooftops.
According to the agency’s press office, eight “Gorodovoy” complexes will begin operating with the first snowfall. Their main task will be to inspect buildings with a history of snow slides or falling icicles.
The neural network was trained throughout the previous winter. Its algorithm learned to recognize snow and ice specifically on St. Petersburg rooftops. Engineers gave the AI a difficult task: distinguishing dangerous icicles from architectural elements such as stucco and bas-reliefs. The system was put into operation only after it achieved high detection accuracy.
Last season, traditional inspections resulted in 21 property owners being fined for failing to clear rooftops on time.
The introduction of artificial intelligence is transforming how the city ensures public safety. It not only improves monitoring efficiency but also serves as a powerful preventive tool. Instead of reacting to incidents, the technology enables authorities to prevent them—ultimately saving lives and protecting public health.
