sgis is a tool that makes it easier identifying PV panels on building roofs. It is based on Qgis and TensorFlow/Keras.

Content

It consists in 3 modules:

- vector_tools: perform very basic operations on vector layers that describes building cadastre data

- splitter: create independant images by intersecting aerial imagery (raster layers) and cadastre data (vector layers)

- classifier: define a convolutional model, have it learn some classification skills and apply this classifier on unlabelled images datasets

Installation

sgis relies on 2 large python packages:

• qgis and its python API
• tensorflow

The recommended way is to set up every dependencies is to use a dedicated Anaconda environment. If needed, please install tensorflow using pip within this environment. Once the environment is created, please clone this repo and install manually using pip within your environment. For instance:

git clone <this_repo>
conda activate my_env
pip install -e /path/to/cloned/repo

Documentation

Please open `doc/build/html/index.html' with a web browser.

Contact

If anything is going wrong or you are looking for more information, please contact Boris Nerot at boris.nerot@univ-smb.fr.