Supervised learning, pivotal in object detection within computer vision, relies heavily on meticulously annotated datasets. Manual annotation, however, is labor-intensive, costly, and prone to human error, hindering scalability. Automated data annotation emerges as a promising solution, leveraging AI-driven tools to enhance efficiency, reduce costs, and ensure consistency. By eliminating subjective biases and fatigue, automated annotation accelerates development and improves the reliability of object detection models. This is particularly crucial in the development of smart intersections, where real-time object detection and tracking are essential for adaptive traffic management and safety. Automated annotation facilitates the efficient processing of large video datasets from intersections, enabling the training of robust models for vehicle, pedestrian, and cyclist detection. This leads to optimized traffic flow, reduced congestion, and improved safety through dynamic signal control and real-time hazard detection.