CV
Junhan Zhu's Curriculum Vitae
Contact Information
| Name | Junhan Zhu |
| Professional Title | Undergraduate Researcher |
| zhujunhan@westlake.edu.cn |
Professional Summary
3rd year undergraduate at Westlake University, researching Computer Vision, Generative AI, and Efficient AI. Actively seeking 27 Fall PhD opportunities.
Experience
-
2026 - Present Nanyang Technological University, Singapore
-
2025 - 2025 Westlake University, Hangzhou, China
Visiting Student
ENCODE Lab, Westlake University
Supervised by Assistant Prof. Huan Wang
- Model Compression and Pruning
- Efficient AI
-
2026 - Present TPAMI Reviewer
Academic Services
Serving as a reviewer for IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI).
Education
-
2026 - 2026 Singapore
Exchange
Nanyang Technological University
Exchange Student
- Exchange student at Nanyang Technological University.
- Research Interests: Computer Vision, Efficient AI, World Model
-
2023 - 2027 (Expected) Hangzhou, China
B.Eng
Westlake University
Electronic and Information Engineering
- Admitted by skipping the final year of high school.
- GPA: 4.04 / 4.3, Rank: 2 / 23
- Research Interests: Computer Vision, Generative AI, Efficient AI
Awards
-
2025 Outstanding Bachelor's Student
Westlake University
Awarded for outstanding academic performance.
-
2024, 2025 Outstanding Undergraduate Scholarship
Westlake University
Awarded for outstanding undergraduate students, 2000 RMB each year.
-
2024, 2025 Innovation Award
Westlake University
Awarded for innovative research.
-
2024 Hongyi Scholarship
Westlake University
Awarded for meaningful social activities.
Publications
-
2026 OBS-Diff: Accurate Pruning For Diffusion Models in One-Shot
ICLR 2026
First training-free, one-shot pruning framework for diffusion models with broad applicability across diverse architectures.
-
2026 Cross-Resolution Diffusion Models via Network Pruning
CVPR 2026 Findings
CR-Diff repurposes network pruning to enhance generalizability by removing adverse weights causing degradation at non-default resolutions.
-
2026 LVOmniBench: Pioneering Long Audio-Video Understanding Evaluation for Omnimodal LLMs
Under Review
A pioneering comprehensive benchmark for evaluating OmniLLMs in Long Audio-Video Understanding.