save_sample_images in train.py

sahilg06 / EmoGen

PyTorch Implementation for Paper "Emotionally Enhanced Talking Face Generation" (ICCVW'23 and ACM-MMW'23)

Creative Commons Zero v1.0 Universal

343 stars 27 forks source link

Closed m00yu closed 11 months ago

m00yu commented 12 months ago

Hello I am training train.py and I am curious what is the meaning of each part in the saved images.

After training, I got results like below. Can you explain about what each part of the image is?

Thank you.

sahilg06 commented 12 months ago

Hi, they are "reference frame || masked frame || generated frame || ground truth frame" respectively.

m00yu commented 12 months ago

Thank you! Can you tell me what is the reference frame?

sahilg06 commented 11 months ago

They are the randomly sampled frames passed as input along with masked frames. You can get more details in the paper itself.