Passengers Monitoring Models at Jakarta Bus Stops Using the YOLOv3

Public transportation plays a crucial role in supporting urban mobility in big cities such as Jakarta. This condition can lead to overcrowding issue at bus stops, especially during working hours, such as in the morning when people go to work or in the evening when people are going home. The objective of this research is to analyze the density pattern at S. Parman Transjakarta bus stop, during the evening, to determine important strategies for addressing the overcrowding issue. The AI model that was proposed and used in this study was based on a YOLOv3 algorithm that was trained on the customized COCO dataset. Our dataset will contain 2000 images that contain at least 10 people, then we will split it into a training set and testing set. This model was utilized for detecting the number of passengers from photographs taken over a span of five consecutive days, from Monday through Friday. The result shows that the number of passengers is not constant and varies by the day of the week. This finding was expected to be a source of knowledge that can help in improving the efficiency of public transportation services as well as the passengers' experience.