The project seeks to explore the possibilities of AI and machine learning in combination with register data to further our understanding of social and sociodemographic segregation and inequality. By using satellite imagery we seeks to redefine how we understand the dynamic neighborhood; we know how poverty and deprivation is measured but we have very little direct understanding of how these phenomena can be identified by visual data. By using a combination of convolutional neural networks and predictive models, we ask the question; how can we better understand inequality at neighborhood level?