Table of Contents:
  • Deep learning backgrounds
  • Software
  • Data used : the Tokyo dataset
  • A simple convolutional neural network
  • Fully convolutional neural network
  • Classifiers on deep features
  • Dealing with multiple sources
  • Semantic segmentation of optical imagery
  • Data used : the Amsterdam dataset
  • Mapping buildings
  • Gap filling of optical images : principle
  • The Marmande dataset
  • Pre-processing
  • Model training
  • Inference.