- Ensure that Anaconda is installed on the computer (it provides Python and Conda)
- Download Nvidia CUDA Toolkit and cuDNN (needed for TensorFlow GPU and any other machine learning library performing calculations on the GPU)
- Download the repository listed above
- Go to the train_data directory and unpack the archive
train_data.zip.001
(it will unpack all archives) - Open a command console in the root of the repository
- Type
conda env create --file environment.yaml
(this will create the python virtual environment and download all required packages) - Type
conda activate tensorflow
(it will activate the environment) - Type
python main.py
(this will run the program)
Step 7 has to be made every time console window is reloaded.