Object Detection Algorithms for Parking Detection - Survey

Abstract

Parking detection plays a pivotal role in the development of smart cities, aiding in the efficient management of urban parking spaces. With the advent of edge computing, devices like the NVIDIA Jetson Nano have emerged as powerful tools for real-time processing in such applications. This research aims to benchmark various object detection algorithms on the Jetson Nano to determine their efficacy and efficiency in parking detection tasks. Traditional and deep learning-based algorithms, including YOLO, Faster R-CNN, and SSD, are being evaluated in terms of accuracy, computational speed, and power consumption. Preliminary results indicate that while deep learning algorithms exhibit high accuracy, their performance varies based on the complexities of the parking environment and the computational constraints of the Jetson Nano. This study provides insights into the optimal deployment of object detection algorithms for parking detection on edge devices, paving the way for the development of cost-effective and efficient smart parking solutions.