virtual_webcam_background icon indicating copy to clipboard operation
virtual_webcam_background copied to clipboard

Published 20 hours ago • verified publisher icon allo-

possible synergies with backscrub

Open coro1404 opened this issue 3 years ago • 1 comments

Hi

just a heads up for a similar project (https://github.com/floe/backscrub) which has implemented the Google Meet Segmentation Model for tensorflow - I can actually achieve 25+ Frames on a normal desktop CPU without GPU and moderate (to almost none) CPU Usage. Could the tensorflow model be adapted easily?

regards, Stephan

coro1404 avatar Nov 30 '21 14:11 coro1404

Aren't they using mediapipe? We have basic support for mediapipe, see #75.

allo- avatar Nov 30 '21 16:11 allo-