Researchers at Istanbul Technical University (ITU) have developed an innovative AI-powered system that can calculate vehicle emissions in real time using traffic camera footage, eliminating the need for fixed air quality monitoring stations.
The project, led by Prof. Dr. Levent Kuzu from ITU’s Department of Environmental Engineering, integrates artificial intelligence with computational fluid dynamics to produce high-resolution air quality predictions for urban areas.
Speaking to environmental news platform Yeşilhat, Kuzu explained that his team analysed images from traffic cameras with deep learning algorithms to determine the type of vehicles and their speed.
Following this, the estimated amount of emissions generated by each vehicle group is calculated using emission factors specific to the vehicle type detected.
Finally, the contribution of these emissions to the ambient air is calculated by taking into account meteorological data, allowing the concentration of pollutants at a given point to be predicted with high accuracy.
The system has been successfully tested using traffic cameras on Istanbul’s Beşiktaş Barbaros Boulevard. In pilot studies, the model accurately identified vehicles over 95% of the time and was able to reliably calculate emission concentrations. The model’s accuracy was validated using data from the air quality station at Yıldız Technical University’s Beşiktaş Campus.
Prof. Kuzu pointed out that the most critical aspect of the system is its ability to determine the number and type of vehicles instantly through image processing. ‘The advantage of our system is that, with just a video feed, emissions can be estimated anywhere – on main roads or side streets – making it highly adaptable for various urban planning and environmental monitoring purposes.’
Last yearn Prof. Kuzu and his team at the University published the research which has led to the Istanbul trial being undertaken. Titled: Development of Artificial Intelligent-Based Methodology to Prepare Input for Estimating Vehicle Emissions, the full research can be read here.
Leave a Reply