/faceswap-GAN

A denoising autoencoder + adversarial losses and attention mechanisms for face swapping.

Primary LanguageJupyter Notebook

faceswap-GAN

Adding Adversarial loss and perceptual loss (VGGface) to deepfakes'(reddit user) auto-encoder architecture.

Updates

Date    Update
2018-08-27     Colab support: A colab notebook for faceswap-GAN v2.2 is provided.
2018-07-25     Data preparation: Add a new notebook for video pre-processing in which MTCNN is used for face detection as well as face alignment.
2018-06-29     Model architecture: faceswap-GAN v2.2 now supports different output resolutions: 64x64, 128x128, and 256x256. Default RESOLUTION = 64 can be changed in the config cell of v2.2 notebook.
2018-06-25     New version: faceswap-GAN v2.2 has been released. The main improvements of v2.2 model are its capability of generating realistic and consistent eye movements (results are shown below, or Ctrl+F for eyes), as well as higher video quality with face alignment.
2018-06-06     Model architecture: Add a self-attention mechanism proposed in SAGAN into V2 GAN model. (Note: There is still no official code release for SAGAN, the implementation in this repo. could be wrong. We'll keep an eye on it.)

Google Colab support

Here is a playground notebook for faceswap-GAN v2.2 on Google Colab. Users can train their own model in the browser.

Descriptions

faceswap-GAN v2.2

  • FaceSwap_GAN_v2.2_train_test.ipynb

    • Notebook for model training of faceswap-GAN model version 2.2.
    • This notebook also provides code for still image transformation at the bottom.
    • Require additional training images generated through prep_binary_masks.ipynb.
  • FaceSwap_GAN_v2.2_video_conversion.ipynb

    • Notebook for video conversion of faceswap-GAN model version 2.2.
    • Face alignment using 5-points landmarks is introduced to video conversion.
  • prep_binary_masks.ipynb

    • Notebook for training data preprocessing. Output binary masks are save in ./binary_masks/faceA_eyes and ./binary_masks/faceB_eyes folders.
    • Require face_alignment package. (An alternative method for generating binary masks (not requiring face_alignment and dlib packages) can be found in MTCNN_video_face_detection_alignment.ipynb.)
  • MTCNN_video_face_detection_alignment.ipynb

    • This notebook performs face detection/alignment on the input video.
    • Detected faces are saved in ./faces/raw_faces and ./faces/aligned_faces for non-aligned/aligned results respectively.
    • Crude eyes binary masks are also generated and saved in ./faces/binary_masks_eyes. These binary masks can serve as a suboptimal alternative to masks generated through prep_binary_masks.ipynb.

Usage

  1. Run MTCNN_video_face_detection_alignment.ipynb to extract faces from videos. Manually move/rename the aligned face images into ./faceA/ or ./faceB/ folders.
  2. Run prep_binary_masks.ipynb to generate binary masks of training images.
    • You can skip this pre-processing step by (1) setting use_bm_eyes=False in the config cell of the train_test notebook, or (2) use low-quality binary masks generated in step 1.
  3. Run FaceSwap_GAN_v2.2_train_test.ipynb to train models.
  4. Run FaceSwap_GAN_v2.2_video_conversion.ipynb to create videos using the trained models in step 3.

Miscellaneous

Training data format

  • Face images are supposed to be in ./faceA/ or ./faceB/ folder for each taeget respectively.
  • Images will be resized to 256x256 during training.

Generative adversarial networks for face swapping

1. Architecture

enc_arch3d

dec_arch3d

dis_arch3d

2. Results

  • Improved output quality: Adversarial loss improves reconstruction quality of generated images. trump_cage

  • Additional results: This image shows 160 random results generated by v2 GAN with self-attention mechanism (image format: source -> mask -> transformed).

  • Consistent eye movements (v2.2 model): Results of the v2.2 model which specializes on eye direcitons are presented below. V2.2 model generates more realistic eyes within shorter training iteations. (Input gifs are created using DeepWarp.)

    • Top row: v2 model; Bottom row: v2.2 model. In column 1, 3, and 5 show input gifs.
    • v2_eb
    • v2.2_eb
  • Evaluations: Evaluations of the output quality on Trump/Cage dataset can be found here.

The Trump/Cage images are obtained from the reddit user deepfakes' project on pastebin.com.

3. Features

  • VGGFace perceptual loss: Perceptual loss improves direction of eyeballs to be more realistic and consistent with input face. It also smoothes out artifacts in the segmentation mask, resulting higher output quality.

  • Attention mask: Model predicts an attention mask that helps on handling occlusion, eliminating artifacts, and producing natrual skin tone. In below are results transforming Hinako Sano (佐野ひなこ) to Emi Takei (武井咲).

    mask1  mask2

    • From left to right: source face, swapped face (before masking), swapped face (after masking).

    mask_vis

    • From left to right: source face, swapped face (after masking), mask heatmap.
  • Configurable input/output resolution (v2.2): The model supports 64x64, 128x128, and 256x256 outupt resolutions.

  • Face tracking/alignment using MTCNN and Kalman filter during video conversion:

    • MTCNN is introduced for more stable detections and reliable face alignment (FA).
    • Kalman filter smoothen the bounding box positions over frames and eliminate jitter on the swapped face.

    dlib_vs_MTCNN
    comp_FA

  • Training schedule: Notebooks for training provide a predefined training schedule. The above Trump/Cage face-swapping are generated by model trained for 21k iters using TOTAL_ITERS = 30000 predefined training schedule.

    • Training tricks: Swapping the decoders in the late stage of training reduces artifacts caused by the extreme facial expressions. E.g., some of the failure cases (of results above) having their mouth open wide are better transformed using this trick.

    self_attn_and_dec_swapping

  • Eyes-aware training: Introduce high reconstruction loss and edge loss around eyes area, which guides the model to generate realistic eyes.

Frequently asked questions and troubleshooting

1. How does it work?

  • The following illustration shows a very high-level and abstract (but not exactly the same) flowchart of the denoising autoencoder algorithm. The objective functions look like this. flow_chart

2. Previews look good, but it does not transform to the output videos?

  • Model performs its full potential when the input images are preprocessed with face alignment methods.
    • readme_note001

Requirements

Acknowledgments

Code borrows from tjwei, eriklindernoren, fchollet, keras-contrib and reddit user deepfakes' project. The generative network is adopted from CycleGAN. Weights and scripts of MTCNN are from FaceNet. Illustrations are from irasutoya.