Automatic Street Parking Space Detection Using Visual Information and Convolutional Neural Networks

Abstract

This paper proposes a unique real-time street parking detection scheme that utilizes visual information and the YOLOv4 convolutional neural network to accurately detect available parking spaces. We also introduce a new video dataset that is captured specifically for this task and is used for training our network. Our network being the first of its kind, successfully detects available street parking spaces. Performance evaluations of our model confirm its efficacy across all types of scenarios.