Find Someone Who: Visual Commonsense Understanding in Human-Centric Grounding