Towards View-Invariant Intersection Recognition from Videos using Deep Network Ensembles | IEEE Conference Publication | IEEE Xplore