fsgan icon indicating copy to clipboard operation
fsgan copied to clipboard

Published 20 hours ago • verified publisher icon YuvalNirkin

Model Inference Speed

Open RAJA-PARIKSHAT opened this issue 2 years ago • 3 comments

I am running image to video face swap and its taking at least five minutes for processing single pair. Although I have enabled gpu in colab but still. Can someone give me their approximate of inference? @YuvalNirkin @SajjadAemmi

RAJA-PARIKSHAT avatar Dec 09 '21 10:12 RAJA-PARIKSHAT

Hi @RAJA-PARIKSHAT How long is your video?

SajjadAemmi avatar Dec 12 '21 06:12 SajjadAemmi

I got the point of length of video, now I am processing a video of 12 seconds but still it takes around 2 minutes on google co laboratory to process it. @SajjadAemmi. Any suggestion you can give if we want to run in real time

RAJA-PARIKSHAT avatar Dec 12 '21 06:12 RAJA-PARIKSHAT

@RAJA-PARIKSHAT this project is not real-time, not at all

SajjadAemmi avatar Jan 15 '22 05:01 SajjadAemmi