eng-tsmith/Thesis

Master Thesis

Dependencies

You need a anaconda or miniconda to use the environment setting. You can install all dependencies by running one of the following commands from the '/config' folder:

# TensorFlow without GPU
conda env create -f cpu_keras.yml

# TensorFlow with GPU
conda env create -f gpu_keras.yml

Or you can manually install the required libraries (see the contents of the environemnt*.yml files) using pip.

Usage

Run the pretrained model

Start up the Udacity self-driving simulator, choose a scene and press the Autonomous Mode button. Then, run the model as follows:

python drive.py model.h5

To train the model

You'll need the data folder which contains the training images.

python model.py