DeepBillboard-based attack recognition and mitigation

Paper

This project provides the tools needed to create a dataset of attack billboards, and to train a model to recognize and neutralize attack billboards.

Example of arbitrary billboard content creation

Setup

It is recommended to use a virtualenv.

This project has been developed using Python 3.8.6.

Additional required libraries can be installed from the provided requirements.txt file.

app/ is the expected working directory.

pip install -r requirements.txt

Usage

Notebook files can be found under app/src/demos. Notebook files expect the working directory to be the demos/ directory.