The One of the main reasons for the traffic on the streets is the lack of information about the availability of parking slots on the streets. Therefore, the need for a new service which informs drivers about the available parking is more than before critical. In smart cities, there are various sensors for data collection that can be used to analyze parking management. Remote sensing is one of the most essential options for monitoring the traffic situation in cities due to the possibility of covering large areas and processing data in a short time, and imaging at the desired time. Imaging using unmanned platforms with the very high spatial resolution is available today. In processing big data, deep learning methods based on a convolutional neural networks are very efficient. The purpose of this study is to develop an architecture for a convolutional neural networks based on encoder-decoder networks to determine the available parking lots on the street and to suggest the nearest parking lot according to the driver's location. For a case study, reference data provided by the International Society of Photogrametry and Remote Sensing from Potsdam, Germany
