Awesome

GANGen-Detection: A Dataset generated by GANs for Generalizable deepfake Detection

The GANGen-Detection dataset can be downloaded from here. The dataset is organized as follows:

GANGen-Detection
├── AttGAN
│   ├── 0_real
│   └── 1_fake
├── BEGAN
│   ├── 0_real
│   └── 1_fake
├── CramerGAN
│   ├── 0_real
│   └── 1_fake
├── InfoMaxGAN
│   ├── 0_real
│   └── 1_fake
├── MMDGAN
│   ├── 0_real
│   └── 1_fake
├── RelGAN
│   ├── 0_real
│   └── 1_fake
├── S3GAN
│   ├── 0_real
│   └── 1_fake
├── SNGAN
│   ├── 0_real
│   └── 1_fake
└── STGAN
    ├── 0_real
    └── 1_fake

Each folder contains 2k real/fake images. The training set can be found in CNNDetection.

Citation

@misc{chuangchuangtan-GANGen-Detection,
  author = {Chuangchuang Tan, Renshuai Tao, Huan Liu, Yao Zhao},
  title = {GANGen-Detection: A Dataset generated by GANs for Generalizable deepfake Detection},
  year = {2024},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/chuangchuangtan/GANGen-Detection}}
}

Acknowledgment

AttGAN, BEGAN, CramerGAN, InfoMaxGAN, MMDGAN, RelGAN, S3GAN, SNGAN, STGAN