FPGA-SoC Based Smart Parking System Using Computer Vision

Abstract

Rapid urban growth and the steady rise in vehicle ownership have created major parking management challenges in cities. Drivers often waste a lot of time looking for parking spaces. This leads to increased traffic, wasted fuel, and more air pollution. Traditional parking systems depend on sensor-based setups, using ultrasonic or infrared sensors in every parking spot. This approach results in high installation and maintenance costs. This paper presents a real-time smart parking system that uses computer vision and FPGA-SoC architecture to efficiently detect parking spot occupancy. The system captures live images of the parking area with a camera and processes the video stream using a lightweight deep learning model like Tiny-YOLO or MobileNet-SSD. Image preprocessing and inference occur faster thanks to the programmable logic of the FPGA, while the ARM processor takes care of slot mapping and communication tasks. The system offers real-time updates on parking availability through a user interface dashboard. Experimental results show that this approach achieves low latency, high detection accuracy, and improved energy efficiency compared to cloud-based or GPU-based solutions. The proposed architecture offers a scalable and cost-effective option for intelligent transportation systems and smart city applications.