search

zhengyuf / STED-gaze

Code for paper 'Self-learning Transformations for Improving Gaze and Head Redirection'
GNU General Public License v3.0
89 stars 23 forks source link

About functional loss #8

Closed san9569 closed 3 years ago

san9569 commented 3 years ago

Hi, I'm very thankful for sharing your code and this awesome project.

The coefficient of functional redirection loss is 20 that you empirically set from the paper. I wonder why you set the coefficient of that to 200 in config/ST-ED.json.

Can you explain for me? Thank you

zhengyuf commented 3 years ago

Hello, thank you for your interest in our paper. Actually, in equation 7 in the main paper, we have

Screenshot 2021-03-04 at 15 12 04

here, L_feature is set to 10. Then, in the final loss equation, L_F is again multiplyed by 20. That's why the lambda for feature loss is 200. Hope that answered your question.