I'm having a couple of thousand clear satellite pictures and would like to extract features out of them, like outdoor swimming pools, solar panels, presence of green space, ... I was wondering if there is one particular way or algorithm to obtain that kind of information?
Swimming pools in satellite images have been studied in this paper: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.471.1233&rep=rep1&type=pdf. It involves crowdsourcing and active learning of a SVM classifier in a sliding window approach. The same approach could also be used for solar panels and green space.