search

wywu / ReenactGAN

[ECCV 2018] ReenactGAN: Learning to Reenact Faces via Boundary Transfer
https://wywu.github.io/projects/ReenactGAN/ReenactGAN.html
197 stars 41 forks source link

Hi ！ How to get the ground-true boundary maps #9

Open wqz960 opened 4 years ago

wqz960 commented 4 years ago

It seems that the training code of the encode network for transforming the face image to boundary maps does not include in this project, you have provided the pretrained model, can you provide me the code for producing the ground-true boundary maps? Thanks!!!

rosebbb commented 4 years ago

I think you can get the boundary map using the pretrained model - v8_net_boundary_detection.pth. You can find the code that turns image into boundary maps in transformer_model.py (in 'init_Bound') or face2boundary2face_model.py ('self.netBoundary' does the job).

wqz960 commented 4 years ago

@rosebbb emmmm. sorry to bother you! can you provide me the script for transforming the ground true landmarks to the boundary maps instead of a test model? Thank you!

rosebbb commented 4 years ago

I don't find such a script in this repository. The model here takes face images as input and outputs boundary maps, which is what you talked about in your first question. If you want to work on landmarks maybe take a look at another paper by the author on https://wywu.github.io/projects/LAB/LAB.html