Training requires a lot of photos and face models need to be more recent to be effective.

And yet my iPhone can recognize my son from the time he was 9 to the time he is 25 with just one shot labeling…