universea / a-practical-facial-landmark-detector Goto Github PK

Implementation of PFLD A Practical Facial Landmark Detector by pytorch.

• WFLW Dataset Download：
  • WFLW Training and Testing images [Google Drive] [Baidu Drive]
• WFLW Face Annotations：
  • WFLW Face Annotations
• Steps：
  • Unzip above two packages and put them on ./data/WFLW/
  • Move ./data/Mirror98.txt to ./data/WFLW/WFLW_annotations
  • Run cd data
  • Run python3 SetPreparation.py

• Training steps：

  • Run tensorboard --logdir=/Your Path/checkpoint/tensorboard &
  • Run python3 train.py -h get usage
  • Run default parms python train.py
  • Checkpoint checkpoint_epoch_x.pth.tarin./checkpoint/snapshot/
  • You can get training log file from ./checkpoint/train.logs

• Testing steps：

  • Run python test.py -h get usage
  • Run default parms python test.py

• Camera realtime show：

  • Run python camera.py

• PFLD
• pytorch
• pytorch-book