Identifying dominant people in meetings from audio-visual sensors