Introduction to Deep Learning in R for the analysis of UAV-based remote sensing data

OpenGeoHub summer school 2020

Tutorial Notebook

The tutorial notebook with all the code and explanations can be found here:

• HTML
• R Markdown source

Installation

To reproduce the tutorial, certain R packages and other dependencies are required (if you don´t want to install them yourself or if you encounter problems during the installation, you might want to use the Docker image, see below).

Install a current version of R and the following packages:

install.packages(c("keras","tfdatasets","mapview","stars","rsample","gdalUtils","purrr", "magick", "jpeg"))

Install miniconda, keras and tensorflow python dependencies:

• reticulate::install_miniconda()
• keras::install_keras()

To check if everything works:

• reticulate::py_config()
• tensorflow::tf_config()
• keras::is_keras_available()

For installing dependencies of rgdal, stars and magick on linux:

• gdal:sudo apt-get install gdal-bin libgdal-dev

• stars: sudo apt-get install libudunits2-dev

• magick: sudo apt-get install libmagick++-dev

When running any tensorflow-related code for the first time, you will probably get a bunch of messages/warnings (some mean-looking red text). Don´t worry about this. This is mainly tensorflow stating that it is not running as fast as it could (this is because we are not using GPU support and have no specific CPU extensions in order to be compatible with as many systems as possible).

Docker

If you are in for using Docker, I´ve prepared a Docker image that includes all software and data for the tutorial, so you can reproduce the tutorial without installing keras etc. on your host system. If you want to use it, download and install Docker Engine/Desktop:

• Linux: https://docs.docker.com/engine/install/ubuntu/#install-using-the-repository
• Windows: https://docs.docker.com/docker-for-windows/install/
• Mac: https://docs.docker.com/docker-for-mac/install/

Then:

• start Docker Engine
• in your terminal (or power shell) type: docker run -d -p 8788:8787 dachro/oghss:latest
  • 8788 on the left side of the : is the port on your host system. You can use any other available port that your prefer
• after the container has been built and started, open you browser and go to localhost:8788/ (or whatever port you have chosen)
• you should see the rstudio server login page. Log in via username: rstudio and password: rstudio
• you are good to go, you don´t need to follow the installations instructions above

Data

Download and unzip the tutorial data from: https://uni-muenster.sciebo.de/s/SOjP7fRGBRztn9z/download