Tenancy Status Identification of Parking Slots Using Mobile Net Binary Classifier

Abstract

The inefficiency in accessing the tenancy status of the parking slots is mainly due to the results of irregular parking regulation/management. The effective parking management enables to avoid unwanted traffic jams and unnecessary fuel wastages. So an efficient parking is necessary for the developing smart cities that aim for a better way of living. So the paper uses the Mobile-Net Classifier to sort out the tenancy state of the parking slots in the cities to assist a proper parking regulation with better proficiency and perfect management. The Mobile-Net classifiers are a sort of light weight deep neural networks that help in identifying the parking slots available accurately based on the image mined from the live camera that feeds the status of the parking lot continuously. The mechanism put forth detects the patches of images form the live recorded video perfectly and determines the vacant slots. The laid out model was applied in an outdoor parking area to determine the systems effective working on the terms of detection accuracy, false positive and the false negative, true positive and the true negative rates along with the average speed of the in identifying the parking slots. As test case, two different mobile-Net network set up were compared to evaluate the swift ness in processing and the perfectness in detecting.