Predict Socio-Economic Status of an Area from Satellite Image Using Deep Learning

There is a profound need of detecting the socioeconomic status and monitoring of remote areas. `Monitoring' here pertains to the development status or the socio-economic status of the area. The main factors considered for such monitoring are structural build(man-made structures), agricultural cover, water resources and roads. In this paper, a model is proposed to predict the development or socio-economic status of areas by analyzing their satellite images using CNN classification and image segmentation techniques. Both supervised and unsupervised approaches are proposed in the paper. The core modules: Preprocessing of images, CNN for classification, Color HeatMap, Contouring and Segmentation incorporated together yield output in the form of a percentage of occurrence of the four factors along with the socio-economic status. The Mappuzzle software was used to collect the dataset of the satellite images using Google's map services and OpenCV library for preprocessing and other image-related operations.