Speaker
Description
Abstract—This review paper focuses on improving AI-based traffic analytics applications in smart cities by integrating emergency response vehicle (ERV) datasets through transfer learning. The research addresses challenges such as the lack of specialized datasets representing ERVs under diverse conditions and the limitations of conventional traffic management systems in prioritizing emergency vehicles. A comprehensive dataset was created, featuring ERVs like ambulances, fire trucks, and police vehicles, captured under varying angles, lighting, and environmental scenarios. The dataset was used to train a YOLOv12 models, leveraging its advanced vehicle detection capabilities for real-time ERV detection and classification. The developed system integrates AI-powered vehicle detection with CCTV infrastructure to enable efficient tracking and prioritization of ERVs. Performance metrics, including precision, recall, and accuracy, demonstrate the system's ability to adapt to diverse traffic conditions, ensuring smoother emergency response operations. The findings highlight the potential of AI-driven solutions in enhancing urban traffic management and emergency response efficiency. Future work aims to expand dataset diversity and improve integration with smart city infrastructures to achieve greater scalability and reliability.
