Extraction and analysis of massive skeletal information from video data of crowded urban locations for understanding implicit gestures of road users.