Xinshuo Weng
Papers
9
Total Citations
673
H-Index
7
About
Xinshuo Weng is a researcher whose work sits at the intersection of 3D perception, multi-object tracking, and autonomous systems. He is best known for establishing influential baselines in 3D Multi-Object Tracking (MOT), a critical capability for autonomous driving and assistive robotics. His 2020 paper, "3D Multi-Object Tracking: A Baseline and New Evaluation Metrics," has accumulated over 486 citations, making it one of the most referenced works in the field and a foundational benchmark that the research community continues to build upon. By emphasizing both accuracy and practical considerations such as computational efficiency and system simplicity, Weng helped redirect community attention toward deployable, real-world solutions rather than purely theoretical advances. Beyond tracking, Weng has made contributions to sequential point cloud forecasting, monocular ground plane estimation, and time-to-collision prediction from video. His work on SPF2 offers a novel inversion of the traditional detect-then-forecast pipeline, while his research on multi-hypothesis tracking addresses error propagation across perception-prediction-planning stacks. He has also explored robot-human interaction through visual-inertial person localization. Across his body of work, Weng demonstrates a consistent drive to make autonomous perception systems more robust, practical, and capable of operating safely in complex real-world environments.
Research Focus
Key Achievements
Top Papers
- 13D Multi-Object Tracking: A Baseline and New Evaluation Metrics486 citations · 2020
- 2A Baseline for 3D Multi-Object Tracking81 citations · 2019
- 33D Multi-Object Tracking: A Baseline and New Evaluation Metrics38 citations · 2019
- 4
- 5MTP: Multi-hypothesis Tracking and Prediction for Reduced Error Propagation15 citations · 2022
- 6
- 7
- 8
- 9