An Imperceptible Anti-forensic Method, Models and Dataset
Perception matters: exploring imperceptible and transferable anti-forensics forGAN-generated fake face imagery detection
This repository contains the code, models and dataset for the project "Perception matters: exploring imperceptible and transferable anti-forensics for GAN-generated fake face imagery detection".
This code has been tested on Ubuntu 16.04 system, with following pre-requisites.
- python >=3.6.10
- PyTorch >=0.4.1
- torchvision >=0.2.1
- Pillow >=5.4.1
The face dataset we used is an image subset dataset downloaded from here.
This face image subset consists of 40,000 real face images and 40,000 fake face images with image resolution as 128x128. For real or fake face images, the dataset splits are: 30,000 images are used for model training; 5,000 images for validation; and the rest 5,000 for test.
After downloading the dataset, please unzip and put them in the data directory.
The pretrained deep learning-based fake face forensic models can be downloaded here. After downloading pretrained models, please uncompress and put them in the checkPoint directory.
cd attacks bash ./run_attack.sh
If you find our code useful in your research, please consider citing our work: