In this paper. a novel Pixel-Voxel network is proposed for dense 3D semantic mapping. which can perform dense 3D mapping while simultaneously recognizing and labelling the semantic category each point in the 3D map. In our approach. https://www.bekindtopets.com/flash-offer-Freetress-Fullcap-Drawstring-Wig-CRETA-GIRL-clearance/