Computer Vision Applications for Smart Cities Using Remote Sensing Data: REVIEW

Abstract

With the rapidly growing urbanization level overpopulation became one of the main challenges for the municipalities, country economics, and governmental management.  Many issues concerning waste management, city resource planning, air pollution, traffic and transportation overload and population health issues challenge existing infrastructures. Smart city aims to improve people's lifestyle, create more sustainable environments and make it easier to manage all these processes by municipalities and governments. The application of AI and computer vision techniques can solve smart city problems (surveillance, area coverage, land usage and land coverage, damage monitoring, fire detection, etc.) that weren't possible or easy to do a few years ago.

Here, we are reviewing deep learning and computer vision applications for smart cities using remote sensing data. Moreover, we present 2 types of data sources for creating datasets to solve smart city problems for cities in Armenia. The first type is potential data sources that can be collected through the efforts of the municipality, the second type is open-source data ready to be used for the solutions below.

First, historical data from 3 different satellites were used to calculate available open-source data that can be used to apply deep learning algorithms. Next, for each type of data and each smart city task, the most convenient deep learning methods were found and described. All techniques were summarized in Table 1.