Satellite images Segmentation Using Unet and VGG16 Architecture

Image segmentation is a common problem in the deep learning field, Satellite image segmentation is one of the most promising applications , using only 16 images in the training set we've achieved state of the art accuracy using transfer learning with VGG16 and the UNET architecture , which proves that UNET achieves very good results without any pretraining , using this network we can count the number of roofs of an entire country combined with the luminosity data , we can can have an idea about the Photovoltaic potential and potential , applications are countless. 1-https://www.linkedin.com/pulse/satellite-images-segmentation-unet-vs-transfer-vgg16-anass-boussarhan/