Extracting semantic video objects | IEEE Journals & Magazine | IEEE Xplore