could someone tell me (as a novice), which software and especially technique (e.g. cnn) is used to realize the Faceswap application and its functionality? You nice guys could really help me out with my presentation!
Thanks a lot.
Greetings from Germany
Zomboid
Re: Process of video processing
Posted: Wed Nov 06, 2019 4:51 pm
by kilroythethird
We are using keras for our models with the tensorflow backend for nvidia cards and the plaidml one for amd cards.
The training guide has a lot of information about how the whole thing works. viewtopic.php?f=6&t=146
Additionally our main devs are at the devoxx conference currently and will hold a talk about faceswap tomorrow: https://devoxx.be/talk/?id=103201
Re: Process of video processing
Posted: Sat Nov 09, 2019 2:12 pm
by bryanlyon
See our talk here:
We answer many questions about how Faceswap works.