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

Parking lots are no longer practical solutions but become another topic of urban mobility problem due to the difficulty in finding available parking spaces. This work proposes a parking space detection system to assist drivers. The system detects unoccupied vacancies by image processing techniques and convolutional neural networks. Vacancies are detected through horizontal markings and by recognizing spaces with or without vehicles. Finally, a mobile application makes available to the user the occupancy status of vacancies. Initial results showed that the system detects vacancies with visible markings during the daytime. To improve detection in adverse situations, the vacancy detection algorithm is being improved.