Learning semantic visual concepts from video | IEEE Conference Publication | IEEE Xplore