Building segmentation with Inception-Unet and classical methods

Remote sensing studies for automatic detection of buildings in urban areas have become important with the increase in spatial resolution of satellite images. Many classical image processing based methods have been proposed in this regard, and deep learning methods have gained popularity recently. In this study, Unet architecture is used with proposed additional segmentation module, and Inception-Unet architecture is proposed to improve Unet performance. The results are compared with classical image processing methods.