Project of Deep Learning in Data science (KTH DD2424):
- Quentin Lemaire
- Quentin Vecchio
- Kevin Yeramian
sudo apt-get update
sudo apt-get install python python3-pip
(You can also usepython2.7
)sudo apt-get install git
git clone https://github.com/ntasfi/PyGame-Learning-Environment
cd PyGame-Learning-Environment
pip install -e . && cd ..
pip install future tensorflow opencv-python
(You can choose betweentensorflow
andtensorflow-gpu
)pip install pygame
python training.py
python play.py
By default, the script play.py
use the old network model. If you want to use your own model (trained with the training.py script) you have to edit the script play.py
to use the Network class and not the NetworkOld class.
This work is based on the code of Yen-Chen Lin (https://github.com/yenchenlin/DeepLearningFlappyBird/blob/master/deep_q_network.py).