Jiaming Sun

孙佳明
suenjiaming@gmail.com
Computer Vision Researcher, SenseTime
B.Eng, Southern University of Science and Technology

I'm currently working at the MIG-3D&AR Group of SenseTime Hangzhou. I'm also affiliated with State Key Lab of CAD & CG, Zhejiang University (ZJU), working closely with the 3D Vision Group lead by Prof. Xiaowei Zhou. Before coming to SenseTime and ZJU, I worked as a research intern in DJI for a wonderful year, mentored by Xiaozhi Chen. I received my bachelor's degree in 2017 from the department of CSE, SUSTech, which is a newly established university in Shenzhen, China (Short story about our university). I worked with Prof. Qi Hao during my undergraduate study.

Research Interests

I have a broad interest in the science and engineering aspects of intelligent systems, with a particular focus on Computer Vision and Robotics. In specific, my interests land on the intersection area between 3D Computer Vision and Deep Learning, with its applications in Augmented Reality and Autonomous Driving. I have been working on several fundamental aspects of 3D Vision, including 3D semantics, 3D representation learning, 2D/3D correspondence estimation, shape reconstruction and pose estimation.

Recently, my research focuses on new SLAM (or SfM and MVS in the offline scenario) pipelines that seamlessly integrate the power of neural networks into the geometric structure of the problem. Going forward, I am particularly interested in improving these perception pipelines to be self-supervised in the context of interactive environments with an intelligent agent.

Recent News

  • 👏 Two papers accepted to CVPR 2021 including one oral. ()
  • 👋 Internship openings ()
    I'm looking for self-motivated interns interested in 3D Vision, Deep Learning and Robotics. Please drop me an email if you believe you would be a good match.

Publications
show selectedshow all by dateshow all by topic

* denotes equal contribution.
NeuralRecon: Real-Time Coherent 3D Reconstruction from Monocular Video
CVPR 2021  Oral Presentation
PaperProject PageCode
LoFTR: Detector-Free Local Feature Matching with Transformers
CVPR 2021  
PaperProject PageCode
Disp R-CNN thumbnail loading...
Disp R-CNN: Stereo 3D Object Detection via Shape Prior Guided Instance Disparity Estimation
CVPR 2020  
PaperCode

Projects

Collaborators