Multi-view deep learning for consistent semantic mapping with RGB-D cameras | IEEE Conference Publication | IEEE Xplore