Is it possible to run this using GPU accelerated machine learning?
The server this is currently running on has an Intel GPU (i9 integrated) and a Google Coral TPU, but when this image runs it puts 200% load on CPU.
This adds ~20W extra consumption (plus heat and noise) on my mini-pc...
Looks like this uses skimage and numpy which are not at all GPU accelerated, which is unfortunate...
Is it possible to run this using GPU accelerated machine learning?
The server this is currently running on has an Intel GPU (i9 integrated) and a Google Coral TPU, but when this image runs it puts 200% load on CPU.
This adds ~20W extra consumption (plus heat and noise) on my mini-pc...
Looks like this uses
skimageandnumpywhich are not at all GPU accelerated, which is unfortunate...