/Distributed-Deep-Fake

Primary LanguagePython

Distributed-Deep-Fake

Install dependencies

https://gist.github.com/zhanwenchen/e520767a409325d9961072f666815bb8

Required pip packages

  • Python 3.6
  • dlib
  • face_recognition
  • numpy
  • keras 2.2.0
  • tensorflow (tensorflow-gpu)

Training images

  • Create data/ folder
  • Inside data folder, create folders A and B to store training images for people A and B.

Run Distributed Version

  • dist_tf_trainer.py
  • Setup network configurations on setClusterConfig function, where ps are parameters servers.

Run Multi-GPU version/Normal version

  • trainer.py
  • Have CUDA setup to Multi-GPU
  • autoencoder.py is setup to detect multi-gpu systems