ki-keras-belgian-traffic-sign

This project was created for the AI course at the University of Pforzheim in the Master's class Embedded Systems. It's goal was to get in touch with artifical intelligence and neural networks based on the Belgian Traffic Sign Dataset. Therefore, the project report as TeX and PDF is also a part of this repository.

Setup project

The following instruction is based on using virtualenv. All paths, used in the source code and documentation, are just written and tested for Linux. May you have to adjust them.

Install requirements

pip install -r requirements.txt

Start training

It is assumed, the data is already downloaded and unpacked in data. If this is not the case, you can comment in the call download_data() at the methode setup_data(). Do not forget to check if the paths are ok for your setup.

python3 Keras.py

Use the trained models

python3 ClassifyImages.py -m <path/to/model> -i <path/to/pic>

Bear the format of the trainings pictures in mind. All models will just predict correct, if the image sections are simmilar to the ones, used for training.