MV2DFusion: Leveraging Modality-Specific Object Semantics for Multi-Modal 3D Detection | IEEE Journals & Magazine | IEEE Xplore