Parking Management System Based on Key Points Detection

In urban areas, efficient parking management is crucial for reducing traffic congestion and environmental impact. This research introduces a new view for making the parking management system that leverages the capabilities of the NVidia Jetson Nano Single Board Computer (SBC) and OpenCV for real-time detection and classification of parking slot occupancy. Unlike traditional systems that rely on intrusive sensors, our proposed solution employs non-intrusive Oriented Fast and Rotated Brief (ORB) key point detection techniques using video feeds. The system architecture integrates video stream processing, ORB via OpenCV, cloud-based data storage, and a Flask server for user notifications. The methodology prioritizes traditional computer vision methods optimized for the Jetson Nano’s CUDA cores, offering a computationally efficient alternative to deep learning approaches. Python’s versatility and MongoDB’s document-based storage are employed for backend development. Our system’s performance, evaluated using open datasets, demonstrates high accuracy, precision, recall, and F1 scores, underlining its effectiveness in real-world urban parking scenarios. This study not only presents a robust solution for parking management but also opens avenues for similar applications in traffic measurement and urban planning.