Tom Bu

I am currently an associate researcher in the Media Analytics team headed by Prof. Manmohan Chandraker at NEC Laboratories America. I work on 3D Computer Vision related to Assisted and Autonomous Driving.

I graduated from the Robotics Institute in the School of Computer Science at Carnegie Mellon University. I was co-advised by Prof. John Dolan and Christoph Mertz. I earned my bachelor’s degree in the School of Engineering and Applied Sciences from Yale University in 2017.

My research interest lies at the intersection of computer vision, machine learning, and robotics. I am working on topcis including domain adaptation, perception for autonomous vehicles, and geometry-based vision. Through my research, I want to create methods that have 3D spatial understanding, stable scene recognition, and the capacity to provide intelligent insights to our dynamic world.

San Jose, CA

Education

M.S in Robotics
School of Computer Science
Aug 2020-Aug 2022

B.S. in Engineering
School of Engineering and Applied Science
Aug 2013-May 2017

Publications

  • Toward Map Updates With Crosswalk Change Detection Using a Monocular Bus Camera
  • Tom Bu, Christoph Mertz, John Dolan
  • Intelligent Vehicle Symposium (IV), 2023
  • Oral Presentation
  • [Paper] [Project] [Code]
  • Toward HD Map Updates With Crosswalk Change Detection From Vehicle-Mounted Cameras
  • Tom Bu, Christoph Mertz, John Dolan
  • Master's Thesis, 2022
  • [Paper] [Poster] [Bibtex]
  • CARLA Simulated Data for Rare Road Object Detection
  • Tom Bu*, Xinhe Zhang*, Christoph Mertz, John Dolan
  • Intelligent Transportation Systems Conference (ITSC), 2021
  • Oral Presentation. * indicates equal contribution
  • [Talk] [Paper] [Code] [Poster] [Bibtex]

---

Projects

  • Dunhuang Image Restoration Using Attention Residual U-Net
  • Tom Bu
  • International Conference on Computer Vision (ICCV) Workshop, 2019
  • 2nd place solution in the Dunhuang Image Restoration Challenge. Invited talk at the e-Heritage workshop
  • [Slides]
  • Conditional Generator Using Nearest Neighbor Pseudo-labelled Segmentation
  • Tom Bu
  • International Conference on Computer Vision (ICCV) Workshop, 2019
  • 3rd place solution in the OpenEDS Eye Generation Challenge. Invited poster presentation at the Eye Tracking for VR and AR workshop
  • [Poster]

---

Industry Experience




---