Skip to content

Latest commit

 

History

History
26 lines (18 loc) · 1.09 KB

README.md

File metadata and controls

26 lines (18 loc) · 1.09 KB

Deepfake-Detection


The Pytorch implemention of Deepfake Detection based on Faceforensics++

We also reprocuded the MesoNet with pytorch version, and you can use the mesonet network in this project.

Install & Requirements

The code has been tested on pytorch=1.3.1 and python 3.6, please refer to requirements.txt for more details.

To install the python packages

python -m pip install -r requiremnets.txt

Usage

To test with videos

python detect_from_video.py --video_path ./videos/003_000.mp4 --model_path ./pretrained_model/df_c0_best.pkl -o ./output --cuda

To test with images

python test_CNN.py -bz 32 --test_list ./data_list/Deepfakes_c0_299.txt --model_path ./pretrained_model/df_c0_best.pkl

To train a model

python train_CNN.py (Please set the arguments after read the code)

License

The provided implementation is strictly for academic purposes only. Should you be interested in using our technology for any commercial use, please feel free to contact us.