Awesome-GANS-and-Deepfakes

A curated list of GAN & Deepfake papers and repositories.

GANs

Tl;dr GANs contain two networks: a generator and a discriminator. The generator creates 'fake' images from noise to fool the discriminator, which attempts to distinguish real images from 'fake' images. Through adversarial training the generator should generate images that closely resemble training images.

Unconditional GANs

• Vanilla GAN: Generative Adversarial Networks, [paper], [github]
• DCGAN: Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks, [paper], [github]
• WGAN: Wasserstein GAN, [paper], [github]
• WGAN-GP: Improved Training of Wasserstein GANs, [paper], [github]
• RGAN: The relativistic discriminator: a key element missing from standard GAN, [paper], [github]

Conditional GANs

• CGAN: Conditional Generative Adversarial Nets, [paper], [github]
• ACGAN: Conditional Image Synthesis With Auxiliary Classifier GANs, [paper], [github]

Image-to-Image Translation

• CycleGAN: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks, [paper], [github]
• StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation, [paper], [github]
• Pix2Pix: Image-to-Image Translation with Conditional Adversarial Nets, [paper], [github]

Volumetric (3D) Generation

• 3DGAN: Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling, [paper], [github]

Applications using GANs

Anime generator

• Towards the Automatic Anime Characters Creation with Generative Adversarial Networks, [paper]
• [Project] Keras-GAN-Animeface-Character, [github]

Interactive Image generation

• Generative Visual Manipulation on the Natural Image Manifold, [paper], [github]
• Neural Photo Editing with Introspective Adversarial Networks, [paper], [github]

3D Object generation

• 3D Shape Induction from 2D Views of Multiple Objects, [paper]
• Parametric 3D Exploration with Stacked Adversarial Networks, [github], [youtube]
• Fully Convolutional Refined Auto-Encoding Generative Adversarial Networks for 3D Multi Object Scenes, [github], [blog]

Super-resolution

• Image super-resolution through deep learning, [github]
• Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network, [paper], [github]
• High-Quality Face Image Super-Resolution Using Conditional Generative Adversarial Networks, [paper]
• Analyzing Perception-Distortion Tradeoff using Enhanced Perceptual Super-resolution Network, [paper], [github]

Image Inpainting (hole filling)

• Context Encoders: Feature Learning by Inpainting, [paper], [github]
• Semantic Image Inpainting with Perceptual and Contextual Losses, [paper], [github]
• Generative Face Completion, [paper], [github]

Image Segmentation

• Vox2Vox: 3D-GAN for Brain Tumor Segmentation, [paper], [github]

Deepfakes

CNN-based Face-swapping

• Fast Face-swap Using Convolutional Neural Networks, [paper], [github]
• DeepFaceLab: A simple, flexible and extensible face swapping framework, [paper], [github]

GAN-based Face-swapping

• Fewshot Face Translation GAN, [github]
• Faceswap-GAN, [github]

Deepfake Detection

CNN-based methods

• MesoNet [paper], [github]
• Detecting Deep-Fake Videos from Phoneme-Viseme Mismatches, [paper]
• Deep Fake Image Detection Based on Pairwise Learning, [paper]

Datasets

• Google Deepfake Detection Dataset
• FaceForensics++ Dataset
• Facebook Deepfake Detection Challenge (DFDC) Dataset
• "SwapMe and Faceswap" dataset
• "Fake Faces in the Wild (FFW) dataset
• Tampered Face (TAMFA) Dataset
• Celeb-DF(v2) Celebrity Deepfake Dataset
• DeeperForensics-1.0
• Diverse Fake Face Dataset (DFFD)