Intersection User Class Detection Using UAV Images and YOLO-Based Algorithms in Smart Cities

The growing population, increasing transport demand and the resulting challenges require the development of advanced intelligent traffic management systems for smart cities. Modern intelligent traffic management requires the integration of advanced deep learning technologies for critical tasks such as monitoring and assessment. Although high-quality data sets are required to train deep learning-based systems, the available data sets in this area are still limited. This study aims to develop a data set and models for traffic analysis using Unmanned Aerial Vehicle (UAV) images and YOLO-based algorithms. For this purpose, a comprehensive data set was created with 17,876 annotated images of intersections with different geometric configurations and of road users. This data set was used to train and test various YOLO (You Only Look Once) models for object recognition. The test results underline both the quality of the data set and the potential of YOLO models in traffic analysis and serve as a valuable reference for future research.