Learning Multi-View Aggregation In the Wild for Large-Scale 3D Semantic Segmentation | IEEE Conference Publication | IEEE Xplore