pjgerrits / ganmapper Goto Github PK

This is the official repo of GANmapper, a building footprint generator using Generative Adversarial Networks

Use environment.yml to create a conda environment for GANmapper

conda env create -f environment.yml
conda activate GANmapper

Predictions can be carried out by running the following sample code. The name of the city depends on the name of each dataset.

python predict.py --dataroot <path to XYZ tile dir> --checkpoints_dir <path to checkpoint> --name <cityname> 

Testing an area in LA:

python predict.py --dataroot datasets/Exp4/LA/Source --checkpoints_dir checkpoints/Exp3 --name LA 

Testing an area in Singapore:

python predict.py --dataroot datasets/Exp4/Singapore/Source --checkpoints_dir checkpoints/Exp3 --name Singapore 

The result will be produced in XYZ directories in ./results/<cityname>/test_latest/images/fake

You can choose to visualise the tiles in QGIS using a local WMTS server.

For example, use the following url and choose Zomm 16 only.

file:///D:/GANmapper//results/Singapore/test_latest/images/fake/{z}/{x}/{y}.png

If you want the output to be in Geojson polygons, use extract.py

  python extract.py <tile_dir>  <out> 

  python extract.py results/Exp4/LA/test_latest/images/fake LA.geojson

Distributed under the MIT License. See LICENSE for more information.

GANmapper is made possible by using the following packages

• PyTorch
• GeoPandas
• Robosat - mask to feature function is borrowed from robosat
• pix2pix - Model Architecture is heavily borrowed from the awesome repo by junyanz