Deep Learning in Historical Architecture Remote Sensing: Automated Historical Courtyard House Recognition in Yazd, Iran

This research paper reports the process and results of a project to automatically classify historical non-historical buildings using airborne satellite imagery. The case study area is center Yazd, most important site in Iran. New computational scientific methods accessibility images have created more opportunities work on automated architecture feature recognition. Building this, convolutional neural network (CNN) main method for classification task project. distinctive features houses Iran are central courtyards. Based this characteristic, objective recognizing labeling as by CNN model. As result, trained model tested validation dataset has an accuracy rate around 98%. In Sum, reported one first works deep learning Iranian efforts use remote sensing techniques courtyard aerial images.