Jointly Recognizing Object Fluents and Tasks in Egocentric Videos | IEEE Conference Publication | IEEE Xplore