Short Bio

I am a Senior Researcher in Microsoft Research Asia. I work on computer vision problems with special focus on single and multiple camera 3D human pose estimation and tracking. I obtained PhD degree from Peking University in 2016. My advisor is Prof. Yizhou Wang. I spent two years (2012 and 2014) at UCLA as an exchange student, working with Prof. Alan L. Yuille. I obtained my Bachelor degree from Software School of Dalian University of Technology.

I have tried to address the critical challenges of 3D human pose estimation from different perspectives: (1) learning low-dimensional representation of 3D poses (sparse coding and activated simplex models) to reduce ambiguity in monocular pose estimation; (2) multi-view image/IMU feature fusion to address occlusion problem (VoxelPose, CrossFusion, IMU-Fusion); (3) graph neural networks to model body joint correlations (LCN, LCN-Journal); (4) limb length constraints (e.g. optimization-based, learning-based); (5) semi-supervised learning from large datasets (Dual Network Learning). I also worked on multi-object tracking (FairMOT)