Jianbiao Mei

Papers

Total Citations

H-Index

About

Jianbiao Mei is a rising researcher in robotics and autonomous driving, with a sharp focus on LiDAR-based perception. His primary contributions lie in **LiDAR Panoptic Segmentation (LPS)**, a critical task that combines semantic and instance segmentation for 3D scene understanding. Mei has pioneered novel frameworks to overcome the limitations of traditional LPS methods. His work, **CenterLPS** (2023, 5 citations), introduces a center-based instance segmentation strategy that eliminates the need for complex offset predictions, offering a more streamlined and effective approach for autonomous systems. Building on this, **PANet** (2023, 3 citations) further advances the field by proposing a sparse instance proposal and aggregation framework, significantly improving performance on large-scale objects and removing dependency on computationally heavy offset branches. Beyond segmentation, Mei has also contributed to **place recognition** with a coarse-to-fine method (2024, 3 citations) that uses attention-guided descriptors and overlap estimation to enhance both accuracy and efficiency. His work is directly applicable to real-world robotics, where reliable scene understanding is paramount. With a growing citation record and innovative solutions to core perception challenges, Jianbiao Mei is establishing himself as a key contributor to the future of autonomous navigation.