Manyuan Zhang 张满园

I am a Staff Researcher at Meituan-M17 (北斗计划), Hong Kong, working on the LongCat series of foundation models. I am currently focused on physical intelligence, including vision-language-action models and world action models.

I received my Ph.D. from MMLab at the Chinese University of Hong Kong, advised by Hongsheng Li and Xiaogang Wang. Previously, I was a researcher at SenseTime. During my six years there, we built the world's best face recognition model at the time, winning the NIST FRVT and ICCV Masked Face Recognition challenges, and the world's best video recognition model, which won the Kinetics-700 track of the ActivityNet Challenge. We also completed exciting projects including DI-star, DI-drive, and SenseMirage.

Manyuan Zhang

News

Earlier news
  • Two papers accepted to ICLR 2026.
  • One paper accepted to EMNLP 2025.
  • One paper accepted to ICCV 2025.
  • I successfully defended my PhD thesis and officially became Dr. Zhang!
  • One paper accepted to CVPR 2025.
  • Two papers accepted to ECCV2024.
  • One paper accepted to SIGGRAPH2024.
  • Two paper accepted to ICCV2023.
  • I pass the PhD candidate test.
  • I am invited to be a reviewer for NIPS2023 and ICLR2023.
  • One paper accepted to CVPR 2023.
  • I am invited to be a reviewer for CVPR2023 and ICCV2023.
  • One paper accepted to ECCV 2022.
  • I am invited to be a reviewer for ECCV2022 and NIPS2022.
  • I am invited to ’智东西’ to give a talk about imitation learning in automatic driving.
  • We win three championships of ICCV 2021 Masked Face Recognition Challenge on glink360k track, unconstrained track and Webface260M track. Code and solutions will be released very soon.
  • We release DI-drive, the decision intelligence platform for autonomous driving simulation. I am responsible for the imitation learning part.
  • One paper accepted to ICCV 2021.
  • We win the championship of NIST FRVT 1:1.
  • We win the championship of NIST FRVT 1:N.
  • We win 2 championships of ActivityNet on the Spatio-temporal Action Localization (AVA) track and the Trimmed Activity Recognition (Kinetics 700) track.
  • One paper accepted to ECCV 2020.
  • We release the X-Temporal for easily implement SOTA video understanding methods with PyTorch on multiple machines and GPUs.
  • One paper accepted to ICCV 2019 LFR workshop.
  • We win the championship of ICCV19 Multi-Moments in Time (MIT) Challenge.
  • We win the championship of ICCV19 Lightweight Face Recognition Challenge.

Tech Report & Projects

Research

Google Scholar

I work on multimodal understanding and generation, from foundation models to reasoning and interaction with the physical world.

* Equal contribution   + Project lead / corresponding author

Stable-MM-R1: Anchoring Multimodal Reasoning Dynamics via Entropy-Guided Stratification

Yimeng Ye, Shuang Chen, Wenxuan Huang, Manyuan Zhang, Kaituo Feng, Zhangquan Chen, Jiayu Chen, Yucheng Zhou, YiCheng Xiao, ZhiYuan Feng, TIANYU SHI

EMNLP · 2026

Large-scale Masked Face Recognition

Manyuan Zhang, Bingqi Ma, Guanglu Song, Yunxiao Wang, Hongsheng Li, Yu Liu

Technical report · 2021Top-1 solution

Experience

Meituan logo
2025 – present
Meituan-M17, Hong KongStaff Researcher · LongCat foundation models
SenseTime logo
2019 – 2025
SenseTime ResearchResearcher / Research Intern · Face recognition, video understanding, and visual generation
Megvii logo
2018 – 2019
Megvii ResearchResearch Intern · Style transfer with Shuaicheng Liu
ByteDance logo
2018
ByteDance AI LabResearch Intern · Large-scale face recognition
More about my earlier work

During six years at SenseTime, I worked on face recognition systems that won the NIST FRVT and ICCV Masked Face Recognition challenges, and video recognition models that won the ActivityNet Kinetics-700 challenge. I worked with Yu Liu and Guanglu Song.

I also helped build DI-star for StarCraft II, worked on reinforcement learning and imitation learning for autonomous driving in DI-drive, and contributed to the text-to-image product SenseMirage.

Education

CUHK emblem
2021 – 2025
The Chinese University of Hong KongPh.D. · Multimedia Laboratory (MMLab)
UESTC emblem
2015 – 2019
University of Electronic Science and Technology of ChinaB.Eng. · Network Engineering

Challenge awards