cnn-semantic-segmentation-satellite-images
The analysis of satellite imagery can provide vital insights in various corporate and governmental application areas. Semantic segmentation of satellite imagery provides dense pixel-wise knowledge, which is necessary to facilitate such analysis. Deep convolutional neural networks are state of the art, when it comes to semantic segmentation of digital images. However, characteristics unique to satellite imagery like their high-resolution, multi-channel input and limited availability of data-sets challenge the designing, parametrizing and training of the neural network model.
