Yuchen Li

李禹陈

PhD Student in Computer Vision

Mohamed bin Zayed University of Artificial Intelligence

Abu Dhabi, United Arab Emirates

I build models that see the world in 3D — and make it move.

I am a final-year PhD student in the Metaverse Lab at MBZUAI, supervised by Hao Li, working on spatially grounded multimodal intelligence. My research builds vision-language and generative models that understand and create 2D, 3D, video, and 4D content. My representative works include Perceptio, UCanDance, PointNeXt, and 3D-CoMPaT.

1,777: citations on Google Scholar
1,681: for PointNeXt (NeurIPS 2022)
ECCV Oral: 3D CoMPaT, 2.7% acceptance
Amazon · Sony AI: industry research internships

Citation counts as of 2026-10-24

Open to research roles starting 2027

Roles: Research scientist, postdoctoral and industry research positions
Topics: Multimodal VLMs · spatial reasoning · world models · generative AI
Start: 2027, after completing my PhD at MBZUAI
Location: United States (on-site or hybrid); currently in Abu Dhabi

yuchen [dot] li [AT] mbzuai [dot] ac [dot] ae

Work Hard, Play Hard, El Psy Congroo!

Education

PhD in Computer Vision
Mohamed bin Zayed University of Artificial Intelligence , 2023 - 2027
M.Sc. in Computer Science
King Abdullah University of Science and Technology , 2020 - 2022
B.Sc. in Computer Science and Technology
Southern University of Science and Technology , 2017 - 2021
Exchange Student, School of Computing
National University of Singapore , 2020

Research Interests

Perceptio

Spatial-token VLM / Amazon Science internship

UCanDance

Music-driven long-horizon video generation / Sony AI

PointNeXt

NeurIPS, 1,650+ citations

3D-CoMPaT

ECCV Oral / dataset

3DCoMPaT++

TPAMI

Recursive Self-Improvement (RSI)

Interested in AI systems that iteratively improve their own capabilities.

Research

01

Spatially Grounded Multimodal VLMs

Developing perception-enhanced vision-language models that generate spatial tokens for complex 2D and 3D spatial understanding.

Perceptio Amazon Science

Spatial Reasoning 2D & 3D understanding

02

Controllable Video and Motion Generation

Studying controllable video synthesis through long-horizon pose diffusion conditioned on music and motion structure.

UCanDance Sony AI

Pose-conditioned Video Music & motion control

03

Scalable 3D Perception

Building scalable point-cloud networks and training strategies for robust 3D object understanding and recognition.

PointNeXt NeurIPS 2022

Point Cloud Understanding 3D recognition

04

Compositional 3D Content and Datasets

Creating compositional 3D datasets and assets that support material, part, and object-level content generation and recognition.

3D-CoMPaT ECCV Oral

Compositional 3D Assets Parts & materials

News

Last updated: October 2026

2026/10: New blog post: Heidelberg Laureate Forum — the hidden paradise I had been looking for, a week with 200 young mathematicians and computer scientists and 20+ laureates.
2026/10: New research note on Perceptio: letting a VLM emit segmentation and depth tokens before it answers.
2026/09: Published a verified bilingual index of all press, papers and social posts about me — 88 items, each with quoted evidence (blog post).
2026/09: Selected as a Young Researcher at the Heidelberg Laureate Forum (HLF) and awarded the Carl-Zeiss-Stiftung Abbe Grant (Top 3%) in Heidelberg, Germany.
2026/09: PointNeXt has surpassed 1,680 citations (1,770+ total citations on Google Scholar).
2026/03: Perceptio, a perception-enhanced VLM for spatial token generation, released as an arXiv preprint.
2025/11: Completed an Applied Scientist Research Internship at Amazon Science, Prime Video.
2025: 3DCoMPaT++ accepted to IEEE TPAMI.
2025: Exploring Scaling Laws of PointNets selected for a 3DV Nectar Track Spotlight Talk.
2024/11: Completed a Research Internship at Sony AI in Tokyo on music-driven dance video generation.
2023/09: Started PhD studies in Computer Vision at MBZUAI.
2023/06: Organized the CVPR 2023 Workshop on Compositional 3D Vision and 3D-CoMPaT Challenge.
2022: PointNeXt accepted to NeurIPS and 3D-CoMPaT accepted to ECCV as an oral presentation.

Featured Publications

Selected publications and preprints. * denotes equal contribution. Full list on Google Scholar.

arXiv26

Perceptio: Perception Enhanced Vision Language Models via Spatial Token Generation

Yuchen Li, Amanmeet Garg, Shalini Chaudhuri, Rui Zhao, Garin Kessler

March 2026 arXiv'26

A perception-enhanced VLM using spatial token generation for 2D and 3D spatial reasoning.

Preprint Project

Under Review

Rethinking Music-Driven Dance Generation in 2D: Long-Horizon Pose Diffusion for Video Synthesis (UCanDance)

Yuchen Li, et al.

January 2026 Under Review

Long-horizon pose diffusion for music-driven dance video synthesis.

NeurIPS22

PointNeXt: Revisiting PointNet++ with Improved Training and Scaling Strategies

Yuchen Li*, Guocheng Qian*, Houwen Peng, Jinjie Mai, Hasan Abed Al Kader Hammoud, Mohamed Elhoseiny, Bernard Ghanem

October 2022 NeurIPS'22

PointNeXt revisits PointNet++ with improved training and scaling strategies for point-cloud understanding.

Preprint Code

ECCV22 Oral

3D CoMPaT Dataset: Composition of Materials on Parts of 3D Things

Yuchen Li*, Ujjwal Upadhyay*, Habib Slim, Tezuesh Varshney, Ahmed Abdelreheem, Arpit Prajapati, Suhail Pothigara, Peter Wonka, Mohamed Elhoseiny

October 2022 ECCV'22 Oral

A large-scale 3D dataset for compositional material and part recognition.

Preprint Code

TPAMI25

3DCoMPaT++: An Improved Large-scale 3D Vision Dataset for Compositional Recognition

Habib Slim, Xiang Li, Yuchen Li, Mahmoud Ahmed, Mohamed Ayman, Ujjwal Upadhyay, Ahmed Abdelreheem, Arpit Prajapati, Suhail Pothigara, Peter Wonka, Mohamed Elhoseiny

January 2025 TPAMI'25

An improved large-scale 3D vision dataset for compositional recognition.

Preprint Code

AAAI Workshop21

Semi-supervised few-shot learning with prototypical random walks

Ahmed Ayyad, Yuchen Li, Raden Muaz, Shadi Albarqouni*, Mohamed Elhoseiny*

January 2021 AAAI Workshop'21

Semi-supervised few-shot learning with prototypical random walks.

Preprint PDF

Experience

Amazon Science - Prime Video

Applied Scientist Research Intern

Seattle, USA

Scaled distributed training for Perceptio, a perception-augmented VLM, using 256 NVIDIA H200 GPUs on AWS for complex 2D and 3D spatial perception reasoning.

Sony AI

Research Intern

Tokyo, Japan

Developed a music-driven dance video generation pipeline and curated high-quality motion datasets for diffusion-based dance generation.

Current

Sep 2023 – Present

Mohamed bin Zayed University of Artificial Intelligence

PhD Researcher, Metaverse Lab (Advisor: Prof. Hao Li)

Abu Dhabi, UAE

PhD research on spatially grounded multimodal intelligence: vision-language models for 2D/3D spatial reasoning (Perceptio) and diffusion-based video and 3D generation (UCanDance).

Dubai Business Associates

Innovation Associate

Dubai, UAE

Selected for a prestigious mini-MBA program under the privilege of Sheikh Mohammed bin Rashid, with a 0.3% acceptance rate.

Emirates Airlines

Strategy Consultant

Dubai, UAE

Consulted for Emirates Airlines on a data-driven AI chatbot and a three-year strategic performance roadmap.

Dubai Future Foundation

Innovation Consultant

Dubai, UAE

Worked on sidewalk delivery robot perception and navigation.

KAUST - Vision CAIR Group

Research Assistant

Thuwal, Saudi Arabia

Worked on semi-supervised few-shot learning, meta-learning, and 3D object recognition.

Alibaba Summer of Code @ Apache RocketMQ

Student Developer

Remote

Developed the Apache RocketMQ JDBC Connector as an independent student developer.

University of British Columbia

Visiting Student, Electronics and Computer Engineering

Vancouver, Canada

Visited the Electronics and Computer Engineering program.

Tsinghua University - Digital Manufacturing Lab

Visiting Research Student

Beijing, China

Led a team to research microstructure prediction modeling of the cladding process using CNNs.

Beyond the Lab

You might know me from Xiaohongshu as the Dubai Prince (迪拜王子). Off the clock I do stand-up, once spent 72 hours with only AI for company, and am slowly buying coffee for a hundred strangers.

Bilibili ↗

Xiaohongshu · DubaiPrince 6.1K followers on Xiaohongshu as “the Dubai Prince” 422 notes on studying, working and living in the Gulf. Follow along ↗ 100 strangers, 100 coffees 56 / 100 coffees with strangers so far — and counting A slow project: one stranger, one coffee, one story at a time. The Hefei stop ↗

Sep 2026 · Heidelberg Heidelberg Laureate Forum: a week with 200 young researchers and 20+ laureates Heidelberg Laureate Forum · Carl-Zeiss-Stiftung Abbe Grant Read the post →

Bilibili ↗

Abu Dhabi · UTC+4 In town? Let's get coffee. Research chats, collaborations, or just a hello. Email me →

Media & Talks

Press coverage

liyc.pw/press · Oct 24, 2026

Verified index: press, papers, talks and social posts

88 items from 31 outlets and platforms (Chinese + English), 2016–2026, each tied to me with quoted evidence.

Browse the index →

36Kr Europe · Jun 12, 2025

72 hours, fully documented: humans' first close encounter with AI In Chinese

Featured in 36Kr's documentary coverage of a 72-hour AI survival challenge, discussing AI agents, AI video creation and human–AI interaction.

Read article ↗

Next Capital · Jan 2024

Dubai Business Associate Li Yuchen on the Middle East 2030 vision In Chinese

Invited by Next Capital to share perspectives on the Middle East 2030 vision and opportunities for Chinese founders in the region.

Read article ↗

Xiaoyuzhou · Podcast

Ep. 07 — Studying in the Middle East: funded by the sheikhs, no tuition, and paid to study! In Chinese

Guest on the DeepMENA podcast, covering KAUST, MBZUAI, Dubai Business Associates, Middle East universities and life across the region.

Listen ↗

Talks & appearances

Sep 2026 Fellow
13th Heidelberg Laureate Forum — Young Researcher & Carl-Zeiss-Stiftung Abbe Grant (Top 3%) Heidelberg Laureate Forum, Heidelberg, Germany
Jun 2025 Stand-up
“I, the Dubai Prince, have shaken hands with a sheikh” — Let's WAVES Gen-Z night 36Kr WAVES 2025, Hangzhou In Chinese
May 2025 Documentary
72-Hour AI Survival Challenge — episode 2, “An AI PhD's 72 hours” 5Y Capital, Shanghai In Chinese
Mar 2025 Spotlight
Exploring Scaling Laws of PointNets — Nectar Track Spotlight Talk International Conference on 3D Vision (3DV 2025)
Jun 2024 Host
“Go to the Middle East! A new option for young Chinese” — sharing session Shenzhen, with Zebra Planet In Chinese
May 2024 Podcast
Ep. 07 — Studying in the Middle East DeepMENA podcast In Chinese
Dec 2023 Talk
The Middle East 2030 vision: opportunities and challenges for Chinese founders Next Capital founder salon #202 In Chinese
Jul 2023 Dialogue
China–Arab Youth Dialogue, as a cultural ambassador Peking University, Beijing
Jun 2023 Workshop
C3DV: 1st Workshop on Compositional 3D Vision — challenge co-organizer CVPR 2023, Vancouver
Oct 2021 Video chat
How do overseas students spend the National Day holiday? — “Houlang Chat Room” TechBeat (Jiangmen Ventures) In Chinese

Community Service

Core organizer: CVPR 2023 Workshop on Compositional 3D Vision and 3D-CoMPaT Challenge.
Reviewer for leading AI conferences and journals, including TPAMI, IJCV, CVPR, ICCV, AAAI, TCSVT, BMVC, TNNLS and NeurIPS.

Awards

2026/09: HLF Young Researcher & Carl-Zeiss-Stiftung Abbe Grant (Top 3%), Heidelberg Laureate Forum & Carl-Zeiss-Stiftung, Heidelberg, Germany.
2023/06: Organizer, CVPR Workshop on Compositional 3D Vision and 3D-CoMPaT Challenge.
2021/05: Super AI Youth, Jiangmen Innovation Ventures.
2019/06: 2nd Prize, ASC World University Student Supercomputer Challenge.
2019/04: Honorable Mention, International Interdisciplinary Contest in Modeling.