I mean, it even recognizes that there are two people in this photo (and one isn't even fully in it), and I can simply click a button to isolate them from the mask. It's a little confused around the child's leg, but fixing that selection is nothing compared to the time spent doing it all manually.