Author name: Chris Choy, Junha Lee, Chunghyun Park, Minsu Cho, Jan Kautz

arXiv:2604.20395v1 Announce Type: cross
Abstract: Open-vocabulary 3D instance segmentation is a core capability for robotics and AR/VR, but prior methods trade one bottleneck for another: multi-stage 2D+3D pipelines aggregate foundation-model outputs …