deepsolar-v2

Reimplementation of DeepSolar using Keras and TensorFlow 2.x

Setup instructions

1. Clone this repository:

   git clone https://github.com/aidan-fitz/deepsolar-v2.git
   

2. Download the dataset from Academic Torrents using a BitTorrent client.
   • Training set: SPI_train.tar.gz
   • Test set: SPI_eval.tar.gz
   • Validation set: SPI_val.tar.gz
3. Extract each of these files into the folder where your repository is:

   cd deepsolar-v2
   tar xzvf SPI_train.tar.gz
   tar xzvf SPI_val.tar.gz
   tar xzvf SPI_eval.tar.gz
   

   Your directory structure should look like this:

   deepsolar-v2/
   |- SPI_train/
   |- SPI_val/
   |- SPI_eval/
   |- [repository contents]
   

4. Create a virtual environment and activate it (requires Python 3.x):

   python3 -m venv env
   source env/bin/activate
   

5. Install the requirements:

   pip3 install -r requirements.txt
   

Running

To run any of the Jupyter notebooks in this repository, run:

jupyter notebook