/face-to-face

face-to-face is a library for easy-to-use human face related trained models.

Primary LanguagePython

face-to-face

Library for easy-to-use human face related trained models.

Models

Face Detection

  • S3FD (FAN)
  • SCRFD

Face Landmark Detection

  • ResNet50d, insightface provided (trained with Microsoft FaceSynthetics)

Face Segmentation

  • BiSeNet

Human matting

  • SGHM (Semantic Guided Human Matting)
  • MODNet

3DMM (3D Morphable Model)

  • BFM
  • FLAME

Face to 3DMM

  • DECA
  • EMOCA
  • MICA
  • Deep3DFaceRecon

Wav2MM (Wav to 3D Morphable Model)

  • FaceFormer

Pretrained Weights

  • Automatically download from release