Detecção de Vagas em Estacionamento Usando Visão Computacional e Redes Neurais Convolucionais

Abstract

Parking lots are no longer practical solutions but become anothertopic of urban mobility problem due to the difficulty in finding availableparking spaces. This work proposes a parking space detectionsystem to assist drivers. The system detects unoccupied vacanciesby image processing techniques and convolutional neural networks.Vacancies are detected through horizontal markings and by recognizingspaces with or without vehicles. Finally, a mobile applicationmakes available to the user the occupancy status of vacancies. Initialresults showed that the system detects vacancies with visiblemarkings during the daytime. To improve detection in adversesituations, the vacancy detection algorithm is being improved.