search

ESanchezLozano / GANnotation

GANnotation (PyTorch): Landmark-guided face to face synthesis using GANs (And a triple consistency loss!)
Other
194 stars 30 forks source link

model.py vs. model_GAN.py #14

Closed YongGeunLee closed 3 years ago

YongGeunLee commented 3 years ago

Generator class in the model.py file in the top-level folder implements "GANimization: Anatomically-aware Facial Animation from a Single Image." (Model to learn by dividing mask and image). However, Generator class in the model_GAN.py file in the Training folder implemented Generator in the basic Encoder-decoder structure, not the idea of the paper. What kind of generator should I use to learn from my custom data?

ESanchezLozano commented 3 years ago

Both are valid, you can replace the training by the one in model.py. The one in model.py corresponds to that of our FG 2020 paper where we presented GANnotation. The one in model_GAN.py in the training subpart is a more recent snippet that works well too.

YongGeunLee commented 3 years ago

Thank you for your reply