Gulnaz Bekkarnayeva

The purpose of the project was to create architect application for semi-automated change detection based on Cognition Network Language in eCognition software. In general, architect application consists of parameter sets, actions, variables, and the rule set. The former serves as a connector between the rule set and actions. In this application user can import, create maps, calculate NDVIs, classify objects, and create links between various possible classes automatically. Next, in segmentation process user can change segmentation scales, reclassify settlements in case of misclassifications, assess the classification accuracy with TTA masks, find statistics of changes based on object links, and export specified classes.