Yu Liu

Yu Liu 

Research Director, SenseTime
PI, Shanghai AI Lab
PhD @ Multimedia Lab, CUHK

liuyuisanai@gmail.com
liuyu@pjlab.org.cn

Google Scholar
Resume
GitHub

News

About me

I am currently working as a research director in SenseTime Group, and also a Principal Investigator in Shanghai AI Lab.

I obtain my PhD from Multimedia Lab (MMLab), the Chinese University of Hong Kong in 2020, supervised by Prof. Xiaogang Wang. Before that I received my bachelor degree in 2016 with best bachelor thesis award. I'm the only awardee of Google PhD Fellowship 2019 in the Greater China area.

My research interests include: Artificial General Intelligence, Deep RL and neural network understanding under billions of data. Previously I was also proficient in basic computer vision topics (detection, classification, recognition, generation, synthesis and video understanding).

In recent years I am a reviewer for CVPR, ICCV, NeurIPS, ICLR, AAAI, IJCAI, T-PAMI, IJCV, TCSVT, TIP, PRCV.

Working Experience

  • PI at Shanghai AI Lab.
    In charge of the Decision Intelligence team and especially the "OpenDILab" open-soursed platform.

  • Research director at SenseTime Research. (since Oct. 2019)
    Leading 3 research departments: X-Lab, Decision Intelligence and Fundamental Model.

  • Research intern at Google Research, Seattle. (Jun 2019 to Oct 2019)
    Worked on neural network understanding and NAS with Research Scientists Xuhui Jia, Mingxing Tan and Raviteja Vemulapalli.

  • Research intern & part-time researcher consultant at SenseTime Research. (Jan 2016 to Jan 2019)
    Worked on large-scale face analysis (detection, recognition, synthesis), generative model and neural network miniaturizing.

  • Research intern at Microsoft Research (Asia), Visual Computing Group. (2015 to 2016)
    Worked on object detection and recognition with Dr. Dong Chen.

Publications

See full list at Google Scholar.
*equal contribition +corresponding author

Projects & Datasets

GitHub

  • OpenDILab, a generalized Decision Intelligence engine.

  • X-Temporal, Easily implement SOTA video understanding methods with PyTorch on multiple machines and GPUs

  • CaffeMex v2.3, a multi-GPU & memory-reduced MAT-Caffe on LINUX and WINDOWS

  • Labeled Pedestrains in the Wild ,a large scale pedestrain re-identification benchmark