Actor and Observer: Joint Modeling of First and Third-Person Videos | IEEE Conference Publication | IEEE Xplore