Deepfake Identification

Identify deepfakes through original and modified XceptionNet models. Feed face images to the network and receive predictions whether the face is real, or a result of deepfake techniques. Built upon pytorch and opencv-python.

Getting Started

It is recommended to use conda for this project. Below are the script to replicate the project environment.

Steps:

1. Create conda environment. below, the env is named torchit, this project is done on python 3.10.
2. Activate the conda env
3. Install packages through pip

conda create -n torchit python=3.10 -y
conda activate torchit
pip install -r requirements.txt

Structure

All codes are available inside src/. Notebooks are available in notebooks/. Scripts in the notebooks expects the data, both raw and preprocessed, to be inside the data/raw and data/preprocessed directory, respectively. Feel free to change accordingly.

License

This python script and its notebooks are licensed under MIT License.