AI system to monitor traffic-related air pollution using cameras

ISTANBUL

Researchers at Istanbul Technical University have developed an artificial intelligence-based method to estimate vehicle emissions using footage from traffic surveillance cameras.

The system, designed by Professor Dr. Levent Kuzu and his team at the university’s Department of Environmental Engineering, combines AI with computational fluid dynamics to forecast traffic-related air pollution in real time.

Funded by TÜBİTAK, Türkiye’s scientific research council, the project generates high-resolution air quality maps by identifying traffic volume and vehicle types through camera feeds.

This allows for more accurate estimates of pollution levels across different parts of a city.

According to Kuzu, the predictions made using this system closely align with observed pollution data.

The process begins with the analysis of traffic camera footage, where AI algorithms classify vehicles, determine their speed and identify their types. Based on this classification, emission levels are calculated using type-specific emission factors.

Finally, these values are used in fluid dynamics models that factor in meteorological conditions to estimate how pollutants disperse through the air. This enables the calculation of pollutant concentration at any given location, using only video data, Kuzu explained.