Animal and Bird Detection
Subject recognition trained on animal anatomy, able to find and hold an eye on a creature rather than a human face.
In depth
The problem is much harder than human faces: the training must cover fur, feathers, wildly different head shapes and heavy camouflage against a cluttered background. Good implementations fall back gracefully — head when the eye is not visible, body when the head is turned away — which is what makes them usable rather than merely impressive.
Examples in practice
Bird detection holds a heron's eye as it lifts from reeds, where a single AF point would have jumped to the vegetation.