Group3D: MLLM-Driven Semantic Grouping for Open-Vocabulary 3D Object Detection Paper • 2603.21944 • Published Mar 23 • 26