aaroncomo / lipfd Goto Github PK

This might be a dumb question but how did you evaluate the algorithm on FF++ which does not have audio.

Thanks for your amazing work! I'd like to ask when your ckpts will be released? Thx!

The loss value needs to be averaged across the time dimension as well as otherwise this will blow up the loss value for RA Loss and overpower the BCE loss. Unless I am missing sth here...

Hi, dear Authors, I see the fantastic work. But how can I download the dataset? And can I use a different dataset like other deep fake datasets or must it be related to the lips or something special?

您好，请问数据集如何获得？

While reading your article, I have a question about how your audio information is embedded in it, as I understand it. The Global Feature Encoder input is just Image, not the spectrogram. Can you ask about the specific processing operation of the input.
Thanks!

thanks for providing such a straightforward technique. I have been doing my thesis and with this help. I admire your work.
I need to use the dataset you provided AVLIP to see how my model is performing on that dataset. i searched everywhere but i could get any of.
I hope you consider my request. thank you.