Home Publications Services Teaching Awards Others

SUN, Pengzhan 孙鹏展

 

SUN, Pengzhan
Ph.D. student


Email: pengzhan AT comp DOT nus DOT edu DOT sg
 
Contact: pengzhansun6 AT gmail DOT com
 
[Google Scholar]

About Me

I'm currently a Ph.D. student at National University of Singapore from 2024 Spring, a member of CVML@NUS.

I received B.S. degree (with Highest Honor) in Mathematics-Physics Fundamental Science from Yingcai Honors College at University of Electronic Science and Technology of China in 2022, supervised by Prof. Wen Li and Prof. Lixin Duan. I received Master of Computing in Artificial Intelligence from School of Computing, National University of Singapore in 2024, supervised by Prof. Angela Yao.

My PhD research follows two connected directions. The first is egocentric assistants: models that understand what a person intends from their first-person view and ground it in the scene, which I study through intention understanding, visual grounding, and VLM post-training. The second is spatial intelligence: models that reason about the 3D physical world from video and plan actions in it, including video spatial reasoning, physical reasoning, and embodied planning. The two meet naturally in wearable AI. An assistant on smart glasses must understand the physical world around its user, such as where things are, how they move, and what can be done with them, yet its purpose is to help that person. Building such assistants is my long-term goal.

Education

 

University of Electronic Science and Technology of China (UESTC), China

Honours Bachelor Degree in Mathematics-Physics Fundamental Science (AI Specialization). Yingcai Honors College. • Sep. 2018 - Jun. 2022
Lab: Data Intelligence Group
Supervisors: Prof. Wen Li and Prof. Lixin Duan    Collaborated with Dr. Bo Wu
 

King Abdullah University of Science and Technology (KAUST), Saudi Arabia

Visiting Student Research Program. Computer, Electrical and Mathematical Sciences and Engineering. • Apr. 2022 - Aug. 2022
Lab: Vision-CAIR
Supervisor: Prof. Mohamed Elhoseiny    Collaborated with Prof. Li Erran Li
 

National University of Singapore (NUS), Singapore

Graduate Student (with Dissertation Option). School of Computing. • Aug. 2022 - Jan. 2024
PhD Student. School of Computing. • Jan. 2024 - Present
Lab: CVML@NUS
Supervisor: Prof. Angela Yao    Collaborated with Dr. Junbin Xiao, Dr. Arjun Akula, and Prof. Linlin Yang

News

Publications

Human-centric Visual Understanding

Understanding people from visual data: egocentric visual intention grounding for assistants, human action and interaction recognition, and human and hand pose estimation.

 

Aligning Thoughts with Answers: Probability Rewards to Tame Thinking Drift

Pengzhan Sun, Shiu-Hong Kao, Shijie Li, Yongyi Su, Junbin Xiao, Arjun Akula, Angela Yao.
arXiv preprint (arXiv), 2026.
[Paper]

 

Dynamic Resolution Routing for Efficient Egocentric Grounding

Huixin Sun, Wangbo Zhao, Fanyue Wei, Qiuxia Lin, Pengzhan Sun, Angela Yao.
Conference on Neural Information Processing Systems (NeurIPS), 2026.
[Paper]

 

Visual Intention Grounding for Egocentric Assistants

Pengzhan Sun, Junbin Xiao, Tze Ho Elden Tse, Yicong Li, Arjun Akula, Angela Yao.
IEEE International Conference on Computer Vision (ICCV), 2025.
[Paper] [Data&Code]

 

Analyzing the Synthetic-to-Real Domain Gap in 3D Hand Pose Estimation

Zhuoran Zhao, Linlin Yang, Pengzhan Sun, Pan Hui, Angela Yao.
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025.
[Paper] [Code]

 

Simultaneous Detection and Interaction Reasoning for Object-Centric Action Recognition

Xunsong Li, Pengzhan Sun, Yangcen Liu, Lixin Duan, Wen Li.
IEEE Transactions on Multimedia (TMM), 2025.
[Paper]

 

Rethinking Visibility in Human Pose Estimation: Occluded Pose Reasoning via Transformers

Pengzhan Sun, Kerui Gu, Yunsong Wang, Linlin Yang, Angela Yao.
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2024 (Oral).
[Paper] [Code] [Video]

 

Counterfactual Debiasing Inference for Compositional Action Recognition.

Pengzhan Sun, Bo Wu, Xunsong Li, Wen Li, Lixin Duan, Chuang Gan.
ACM International Conference on Multimedia (ACM MM), 2021.
[Paper] [Code] [Poster] [Slideshow]

Spatial, Physical & Embodied Intelligence

Reasoning about and acting in the physical world: video spatial reasoning, physical reasoning from video, reliable spatial benchmarks, and embodied planning in simulated and video-derived environments.

 

From Reasoning Failures to Composable Video Spatial Intelligence

Pengzhan Sun, Junbin Xiao, Ramanathan Rajaraman, Shiu-Hong Kao, Angela Yao.
arXiv preprint (arXiv), 2026.
[Paper]

 

ReSTI: A Source-Grounded Audit and Repair of STI-Bench

Pengzhan Sun, Ramanathan Rajaraman, Shiu-Hong Kao, Junbin Xiao, Angela Yao.
arXiv preprint (arXiv), 2026.
[Paper] [Data&Code]

 

Ego2World: Compiling Egocentric Cooking Videos into Executable Worlds for Belief-State Planning

Qinchuan Cheng, Zhantao Gong, Pengzhan Sun, Angela Yao, Shijie Li.
arXiv preprint (arXiv), 2026.
[Paper]

 

PRISM: Planning and Reasoning with Intent in Simulated Embodied Environments

Yunn Kang Lim, Pengzhan Sun, Ziyi Bai, Xun Xu, Angela Yao, Xulei Yang, Shijie Li.
arXiv preprint (arXiv), 2026.
[Paper] [Website]

 

FrameBind: Directional Grounding for Allocentric Spatial Reasoning with VLMs

Ramanathan Rajaraman, Pengzhan Sun, Angela Yao.
Under Review, 2026.
 

CoPhyBench: Benchmarking Physical Reasoning from Conditional Video Observation

Fanyue Wei, Kai Xu, Yizhuo Zhang, Pengzhan Sun, Junbin Xiao, Angela Yao.
Under Review, 2026.
[Paper]

Misc.

 

Rethinking Probability-Based Reinforcement Learning From Posterior Concentration

Shiu-Hong Kao, Yubo Zhao, Zhenyu Tian, Pengzhan Sun, Yicong Li, Angela Yao.
arXiv preprint (arXiv), 2026.
[Paper]

 

ImageCaptioner2: Image Captioner for Image Captioning Bias Amplification Assessment

Eslam Mohamed Bakr, Pengzhan Sun, Li Erran Li, Mohamed Elhoseiny.
AAAI Conference on Artificial Intelligence (AAAI), 2024.
[Paper] [Website]

 

HRS-Bench: Holistic, Reliable and Scalable Benchmark for Text-to-Image Models

Eslam Mohamed Bakr, Pengzhan Sun, Xiaoqian Shen, Faizan Farooq Khan, Li Erran Li, Mohamed Elhoseiny.
IEEE International Conference on Computer Vision (ICCV), 2023.
[Paper] [Code] [Website]

Services

Conference Reviewer:

Journal Reviewer:

  • International Journal of Computer Vision (IJCV)
  • IEEE Transactions on Image Processing (T-IP)

Teaching

Awards

Others