PeopleofSuperintelligence

Researcher

Zicheng Liu

Also known as Liu Zicheng

Zicheng Liu is a computer-vision and vision-language researcher whose work spans 3D body tracking, image and video understanding, and foundation models. He earned degrees in mathematics, operations research, and computer science in China and at Princeton, then joined Microsoft Research in 1997 after work at Silicon Graphics. His Microsoft work includes Azure Kinect body tracking, GIT, and MM-Vid, and IEEE named him a Fellow in 2015. As of October 2026, he is senior director of GenAI at AMD and leads a team for large foundation model training, with recent work on Instella-VL and Instella-MoE.

Works
301
Citations
21,467
h-index
68
Badges

Publication numbers from OpenAlex, as of 18 October 2026.

Timeline

  1. 1984–1996

    Degrees in China and Princeton

    He earns a B.S. in mathematics from Huazhong Normal University in 1984, an M.S. in operations research from the Institute of Applied Mathematics, Chinese Academy of Sciences in 1989, and a computer science Ph.D. from Princeton University in 1996.

  2. 1997

    From Silicon Graphics to Microsoft

    He joins Microsoft Research in 1997. At Silicon Graphics, he develops a trimmed NURBS tessellator shipped in OpenGL and OpenGL Optimizer.

  3. 2015

    An IEEE Fellowship

    IEEE names him a Fellow in 2015 for contributions to visual processing for multimedia interaction.

  4. 2019

    Body tracking with Azure Kinect

    In September 2019, he presents the Azure Kinect Body Tracking SDK, which his team develops as a neural-network solution for 3D skeletal tracking with an RGB-D sensor.

  5. May 2022

    A transformer for vision and language

    In May 2022, he co-authors GIT, a transformer with one image encoder and one text decoder for image and video captioning and question answering. It establishes new results on 12 benchmarks and surpasses human performance on TextCaps.

  6. 2023

    Understanding long videos

    In 2023, he co-authors MM-Vid, a system that combines GPT-4V with vision, audio, and speech tools to analyze long videos and generate scripts for language models.

  7. March 2025

    A vision-language model at AMD

    In March 2025, he co-authors Instella-VL-1B, AMD's first vision-language model for image understanding, trained on MI300X GPUs.

  8. As of October 2026

    Leading AMD's GenAI group

    In July 2026, he co-authors Instella-MoE, a fully open mixture-of-experts language model with 16 billion total parameters and 2.8 billion active parameters. As of October 2026, he is senior director of GenAI at AMD and leads a team for large foundation model training.

A language model matched the links in this timeline to people, organizations and papers (provisional, model-judged).

Career

  1. Since 2024Advanced Micro DevicesSenior director of GenAI
  2. 2025–2026Guizhou University of Finance and Economics
  3. 2022–2026Westlake University
  4. 2020–2022University of Washington
  5. 2000–2024Microsoft
  6. 1997–2005Microsoft Research

Education

  1. 1996Princeton UniversityPhD in computer science
  2. 1989Chinese Academy of SciencesM.S. in operations research
  3. 1984Huazhong Normal UniversityB.S. in mathematics

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. Mining actionlet ensemble for action recognition with depth cameras2012 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 1,560 citationsWith Junsong Yuan and Ying Wu
  2. Dynamic Convolution: Attention Over Convolution Kernels2020 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 1,543 citations
  3. Action recognition based on a bag of 3D points2010 · 2009 IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops · 1,428 citationsWith Zhengyou Zhang
  4. HON4D: Histogram of Oriented 4D Normals for Activity Recognition from Depth Sequences2013 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 956 citations
  5. Rethinking Classification and Localization for Object Detection2020 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 759 citationsWith Yun Raymond Fu

From OpenAlex, as of 18 October 2026.

  1. The company share of AI researchers’ top papers rose from 2% to 33%Includes Microsoft. We estimate where leading AI researchers worked when their most-cited papers appeared. The company share rose from 2% in 1975–1979 to 33% in 2020–2024.
  2. Google has the highest combined h-index among AI organizationsIncludes Microsoft. Our estimates of combined and median h-index at AI organizations, using current affiliations and OpenAlex records.
  3. Google alumni founded or led 77 other organizations, the most among the labs we compareIncludes Microsoft. We trace 177 organizations founded or led by AI lab alumni, and 186 career moves between the labs.
  4. Google had the most people shaping AI every year from 2014 to 2025Includes Microsoft. Our estimates of which organizations employed the most people shaping AI from 1962 to 2025, with a year-by-year chart.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Zicheng Liu yet. Send us a free-licensed photo.

  1. Zicheng Liu | IEEE CASS ieee-cas.org. Checked 18 October 2026.
  2. HOME | Mysite zicliu.wixsite.com. Checked 18 October 2026.
  3. 51 SPS Members Elevated to Fellow | IEEE Signal Processing Society signalprocessingsociety.org. Checked 18 October 2026.
  4. ICIP 2019 | 2019 IEEE International Conference on Image Processing - Taipei, Taiwan - 22-29 September 2019 2019.ieeeicip.org. Checked 18 October 2026.
  5. GIT: A Generative Image-to-text Transformer for Vision and Language - Microsoft Research microsoft.com. Checked 18 October 2026.
  6. MM-Vid: Advancing Video Understanding with GPT-4V(ision) multimodal-vid.github.io. Checked 18 October 2026.
  7. Zicheng Liu — ROCm Blogs rocm.blogs.amd.com. Checked 18 October 2026.
  8. Introducing Instella-MoE: A State-of-the-Art Fully Open Mixture-of-Experts Language Model rocm.blogs.amd.com. Checked 18 October 2026.