Open muyustan opened 4 years ago
the process is quick on my 1080 gpu when model=cnn
the process is quick on my 1080 gpu when model=cnn
Is it using GPU instead of CPU ?
the process is quick on my 1080 gpu when model=cnn
Is it using GPU instead of CPU ?
yes, you need a GPU when set model=cnn
I had the same problem. Change your model to hog which uses ur cpu instead. Its still not great for me but its only a 2 to 3 second delay.
Description
I have a code which works nice, so there should not be any mistakes, the only problem is, it is too slow. The processing of a frame(either an image in unknown_faces dir or video feed) takes like 20-30 seconds.
What I Did
I am getting the image from the video feed(my laptop webcam, nothing fancy).
I insterted some
print()
s to determine at where exactly the code slows down and I found that:locations = face_recognition.face_locations(image, model=MODEL)
the delay occurs during the execution of this line.
Is this normal?
MODEL = "cnn"
, by the way.