Cloud-based automatic building and road extraction from large scale open geospatial datasets
We author Jupyter notebooks to develop deep learning models on Amazon SageMaker instance. These models automatically extract building footprints and road networks from open geospatial datasets. The notebooks reproduce winning algorithms from the SpaceNet challenges. In addition to the SpaceNet satellite images, we introduce USGS 3D Elevation Program (3DEP) light detection and ranging (LiDAR) data to the workflow. We demonstrate using satellite images, LiDAR data, or combination of both to train and test deep learning models for building and road extraction. Both datasets are hosted on Amazon Web Services (AWS). This tutorial will share the notebooks and provide hands-on and step-by-step instructions on running machine learning services to extract features from large scale geospatial data on AWS. At the end of the tutorial, audiences can reproduce the building and road extraction tasks, apply the models to other area of interests where satellite or LiDAR data are available, and innovate with new ideas to improve the performances. The audiences can also appreciate the benefits of cloud computing and storage first-hand.