This repository is for the course Urban computing at the Leiden University - LIACS
the \models directory holds the three trained models from the ConvLSTM architecture.
the STAR.ipynb holds the following:
Implementation of the Spatio-temporal Autoregressive model.
Creation of the gifs that can be found in the \figures folder.
Some more data visualisation.
The Data exploration and neural net implementation.ipynb notebook holds the following:
Data exploration of the Burned area patch dataset
Data exploration of the Burned area pixel dataset
Visualisation of the Burned area pixel dataset
Data pre-processing, model architecture and training of the ConvLSTM models
Visualisations of the results of the trained models.
Make sure to import all necessary packages using pip install -r requirements.txtbefore running anything.
Also, the datasets can be downloaded here:
https://climate.esa.int/es/odp/#/project/fire
Be warned, it is a lot of data and it takes a long time to download everything. The notebooks will explain the folder structure for the dataset creation but this could also be adapted by the user to fit their needs of course.