The increasing burden of vehicular air pollution in urban environments has necessitated the evolution of emission inventory models, particularly through the integration of Artificial Intelligence (AI) techniques. This systematic review investigates past and current research focused on the development and application of AI-driven emission inventory models, with a special emphasis on image-based air pollution detection and vehicle classification in congested traffic areas. The review compiles and analyzes over 25 peer-reviewed articles, technical reports, and case studies published between 2009 and 2024, highlighting the use of machine learning, computer vision, and deep learning techniques to estimate pollutant emissions such as PM2.5, NOx, CO, and VOCs in metropolitan cities. Particular attention is given to methodologies that use traffic camera images, drone footage, and surveillance systems for real-time detection and classification of vehicle types and traffic density, serving as proxies for emission estimates. The study identifies major gaps in spatial-temporal resolution, data validation techniques, and integration with official emission inventories. Finally, it offers future research directions including hybrid models combining AI and traditional inventory methods, heat mapping in urban environments, city-specific calibration, and policy-level applications. This review supports the foundation for advanced, real-time, and scalable emission modeling tools tailored for smart city air quality management.
Kayarwar et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: