Viola-Jones face detection and features per weak classifeirs

Question

From what I've read, each weak classifier in Viola-Jones face detection gets all N training samples and only 1 feature.

1) Does that mean that I will have to check over 160k classifiers at first for 24x24 frame, since there are that many Haar features?
2) Does each weak classifier check it's own feature in each of the N training samples? I. e. uses the fraction of each of N images and checks integral image to compute whether it's face or not?

Answer 1

I've found answers to those questions:

1. Initially yes, during the training the cascade we have to check a lot of frames. This is also why e.g. OpenCV provides pre-trained weights for certain tasks.
2. Yes.