The Performance of Vacant Parking Spaces Detection Using YOLOv5

Parking challenges are increasingly significant in urban areas, particularly in nations with high numbers of private vehicles, such as Indonesia. Conventional parking systems often struggle to effectively identify available spaces, leading to wasted time, congestion, and environmental issues. This study addresses these challenges by implementing YOLOv5, a deep learning-based object detection model, to improve the efficiency and accuracy of vacant parking space detection. The proposed system processes video input from parking areas to detect vehicle presence and determine the occupancy status of designated parking spaces. A dataset of 3,843 images was compiled for model training and testing, encompassing diverse conditions to enhance detection robustness. The model's performance was tested on two video samples recorded under different environmental conditions, achieving a high accuracy rate of 96.4% in detecting vacant and occupied spaces. These results underscore the potential of YOLOv5 for optimizing parking management, reducing the limitations of traditional systems, and providing a scalable solution for smart city infrastructure.