gis_ml_workflow_example

The above notebook provides an example of how to work with raster & vector files using some of the most popular open source libraries such as geopandas, shapely, rasterio etc...

In addition to interacting with raster files, the notebook will demonstrate how to use Solaris to carry out semantic segmentation on the buildings in the raster file.

The above is a work in progress and will be updated to include more examples such as:

• Creating chips from a larger raster image.
• Finetuning a model
• Creating a model from scratch

Solaris Building Segmentation Output

Using Conda

conda create env -n env_name -f environment_solaris.yml python 3.8.6

conda activate env_name

Using Docker

Navigate to the directory containing the docker-compose.yml file: docker_solaris/

Run the commands below in your terminal

docker-compose up -d

docker-compose logs

Extract the notebook address from the command above and run it in the environment of your choice.

Open the solaris_building_segmentation_output_example.ipynb

Cresi Road Segmentation Output

Using Docker

Navigate to the directory containing the docker-compose,yml file: docker_cresi/

Run the commands below in your terminal

docker-compose up -d

docker-compose logs

Extract the notebook address from the command above and run it in the environment of your choice.

• Place the Ortho_Sample.tif file inside the data\cresi directory.
• Place the output csv file wkt_vectors.csv that was generated by running the cresi baseline model on Ortho_Sample.tif inside of the data\cresi directory.

Open and run the cresi_output_example.ipynb

An map.html file should be generated inside of the data\cresi directory.

Open the map.html file in a browser to visualize your output using leaflet.js