How to select anchor points for affine transformations in an image?

Question

Machine learning

Good day, let's say there is an image of a face, but the face in the photo is turned to the side or tilted
. Accordingly, you need to center the face, I plan to use affine transformations, but for them you need to mark a certain number of points on the face
. As I understand it, I need several hundred images with already marked points, then train a neuron on these images so that the input is a regular image, and the output is an image with built points
. But I don’t really understand how to mark these points on the image, draw them in paint or how?
And the second thing, which I don’t really understand, since there is not much experience in this topic, what type of neuron is better to use to learn how to build such points?
help me please