search

HSinger04 / VOGUE-Reimplementation

A reimplementation of the VOGUE paper for the IANNwTF WS20/21 course at the University of Osnabrück
1 stars 0 forks source link

Project images to latent vector z #8

Open HSinger04 opened 3 years ago

HSinger04 commented 3 years ago
  1. Create a CNN that maps from image to latent vector z.
  2. Train it by minimizing the perceptual loss between input and output by StlyeGAN (unclear: perceptual loss)