RDIR: Capturing Temporally-Invariant Representations of Multiple Objects in Videos | IEEE Conference Publication | IEEE Xplore