how does computer vision use machine learning?

I'm reading Szeliski's book about computer vision, and I can't always see a connection between computer vision and machine learning in many places. Especially in instance recognition. Does it exist?

is there any machine learning going on in instance recognition? Is SIFT the only thing required to accomplish this? Does any learning take place?

Is there a paper or other document explaining how a machine learning approach is used in finding objects in an image?