How are faces or objects detected in high resolution video (images), such as 4k (12mp)?

Question

Neural networks

Hello.
I am quite a beginner and understand almost nothing in the detection of formations/faces.

I was told that if I take a model at 300x300px, then it makes no sense to take a video stream from a camera of 1920x1080 or more to work with a 300px model.

And I immediately had a question, how then does the detection of images and faces work on 4k cameras, where there are 8 and 12 megapixels each?

If someone throws a link to some kind of manual where everything is explained on the fingers, I will be very grateful. Thank you.