YOLOv5-based Parking Space Detection for Reducing Urban Congestion and Fuel Wastage

Urban congestion and fuel inefficiency are significant challenges in contemporary cities, often aggravated
by suboptimal parking space management. This research
introduces a high-efficiency YOLOv5-based system for parking
spot identification aimed at delivering real-time availability notifications, decreasing search duration, and reducing fuel usage. The solution combines IoT-enabled cameras with YOLOv5, using its advanced object recognition skills to precisely identify unoccupied and occupied parking spots. Comprehensive testing was performed on parking images from various metropolitan settings under differing lighting and meteorological conditions. The proposed system achieved
accuracy of 99.31%, markedly exceeding traditional image
processing techniques and other deep learning (DL) models.
Real-time detection technology decreased the average parking
search duration and reduction in fuel consumption per car each trip. Moreover, the model exhibited an inference speed of 9 ms per frame, facilitating smooth real-world implementation. These findings highlight the potential of YOLOv5 to revolutionize smart city parking systems, alleviate traffic congestion, and enhance environmental sustainability. Future research will focus on enhancing computational efficiency for extensive urban applications and using predictive analytics for adaptive parking management