The field of 3D computer vision is rapidly advancing, with a focus on improving the efficiency and accuracy of 3D segmentation and tracking methods. Recent developments have seen a shift towards more lightweight and flexible approaches, enabling real-time inference and improved performance in complex environments. Notably, the use of superpoint-based pipelines and visual-spatial tracking is becoming increasingly popular, allowing for more accurate and generalizable results. These advancements have significant implications for applications in robotics, AR/VR, and autonomous driving. Some noteworthy papers include: EZ-SP, which proposes a fast and lightweight superpoint-based 3D segmentation method, achieving state-of-the-art accuracy with significantly reduced computational requirements. S2AM3D, which introduces a scale-controllable part segmentation method, enabling real-time adjustment of segmentation granularity and achieving leading performance across multiple evaluation settings. OpenTrack3D, which presents a generalizable and accurate framework for open-vocabulary 3D instance segmentation, leveraging a novel visual-spatial tracker and multi-modal large language model to enhance performance and compositional reasoning.