Parking on the street Parking Meter Transaction-Based Occupancy

Abstract

In addition to being a difficult undertaking for the majority of drivers, driving in search of parking also contributes to increased congestion and pollution. As a result, researchers and city authorities are showing a growing interest in intelligent parking assistance systems. A significant portion of these systems are dependent on technology that is both costly and not easily scalable, such as real-time parking sensors or video systems. The purpose of this work is to offer a deep learning architecture that, based on digital metre payment events, can make predictions about the current number of cars parked at various locations. We achieve better results than simple baseline models and a probabilistic technique that is considered to be state-of-the-art in the literature. A direct correlation between transactional data and parking occupancy cannot be established because not all individuals adhere to the length or pay that they have paid for. As a result, we will discuss the dependability of our method on a variety of datasets and spatial granularities. In spite of the fact that our model is not as trustworthy as sensor data, particularly for parking zones that are relatively small, our methodology offers a cost-effective approach to infer the occupancy of on-street parking spaces and enables meaningful autonomous parking services.