Effects of Different Optimizations at Detection of Vehicles and Pedestrians with YOLOv8

Improvements in deep learning has facilitated traffic management by providing analyzing and prediction capabilities on the road. In this study, vehicle and pedestrian detection with YOLOv8 was carried out with different optimizations such as Adam, Stochastic Gradient Descent (SGD) and RMSprop, at the end of the experiments, the model with SGD which produced $87.1 \%$ in mAP was seen as the best optimizer when compared with others at detecting objects.