Deep Satellite Resources

List of labelled datasets, projects, and papers relevant to classifying satellite imagery using deep learning.

• Datasets

  • SpaceNet
    • Object Detection on SpaceNet
  • SAT-4 and SAT-6
  • Brazilian Coffee Scenes Dataset
  • Kaggle: Dstl Satellite Imagery
  • Deep learning for Earth Observation

• Projects

  • Project Using OSM Roads

• Papers

  • DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs

  • SEMANTIC SEGMENTATION OF AERIAL IMAGES WITH AN ENSEMBLE OF CNNS

  • A novel spectral index to automatically extract road networks from WorldView-2 satellite imagery

  • Classification and Segmentation of Satellite Orthoimagery Using Convolutional Neural Networks

  • Scene Classification via a Gradient Boosting Random Convolutional Network Framework

  • Semantic Segmentation of Earth Observation Data Using Multimodal and Multi-scale Deep Networks