International Journal of Scientific Research and Engineering Development

Urbanization has enhanced vehicle density, which causes a high traffic jams and slow emergency service. The conventional methods of traffic monitoring are based on the statistical models and the previous congestion data, which are not real-time situational. In this study, a Traffic Vision System to perform Live Urban Traffic Analysis in real-time using deep learning and aerial video is proposed. The suggested system uses YOLO11x-OBB (Oriented Bounding Box) to detect the vehicles precisely in the aerial traffic environment. Video frames are removed with the help of OpenCV, after which the real-time detection, vehicle tracking, and counting of cars with unique vehicles are performed. The system is trained on a personal Emergency Vehicles dataset of 1,366 images and 1,469 annotations of three categories: emergency vehicles, persons and non-emergency vehicles. The experimental outcomes prove that the suggested vision-based vision-based approach gives much better real-time traffic monitoring and emergency response efficiency in comparison with the traditional econometric traffic congestion models. The system has good performance in detection in different climatic conditions and the visual results can be interpreted to be used in traffic monitoring and emergency management systems.

Shifa Bilal Tamboli, Simeen Phiroj Mulani, Arman Tajuddin Shiakh,"Traffic Vision System for Real-Time Urban Traffic Analysis Using Deep Learning" International Journal of Scientific Research and Engineering Development, V9(2): Page(2462-2467) Mar-Apr 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.