I am a Ph.D. student at the University of Sydney, supervised by A/Prof. Chang Xu. Before joining USYD, I spent one year at the City University of Hong Kong as a Research Assistant under the supervision of A/Prof. Minjing Dong. I received my Master’s degree from Tianjin University, where I was advised by Prof. Xiaojie Guo, and my Bachelor’s degree from Central South University.

My research focuses on efficient and generalizable visual perception, including Multimodal Large Language Models (MLLMs) with an emphasis on fine-grained perception and efficient vision-language-action models, State Space Models (e.g., Mamba) for vision, Vision Foundation Models for training-free open-vocabulary segmentation, and object detection in images and videos.

🔥 News

• 2026.04:  📄 New preprint “Q-Zoom: Query-Aware Adaptive Perception for Efficient Multimodal Large Language Models” is on arXiv.
• 2026.01:  🎉 Two papers accepted to ICLR 2026!
• 2026.01:  🎉 Our paper on practical video object detection is accepted by IJCV 2026.
• 2025.09:  I started my PhD at the University of Sydney.
• 2025.07:  🎉 Two papers accepted to ICCV 2025.
• 2024.09:  🎉 “Multi-Scale VMamba” accepted to NeurIPS 2024.

📝 Publications

^† denotes equal contribution.

Catching the Details: Self-Distilled RoI Predictors for Fine-Grained MLLM Perception

Yuheng Shi, Xiaohuan Pei, Minjing Dong, Chang Xu

[Code]

International Conference on Learning Representations (ICLR), 2026

Practical Video Object Detection via Feature Selection and Aggregation

Yuheng Shi^†, Tong Zhang^†, Xiaojie Guo

[Code]

International Journal of Computer Vision (IJCV), 2026

VSSD: Vision Mamba with Non-Causal State Space Duality

Yuheng Shi, Minjing Dong, Mingjia Li, Chang Xu

[Code]

IEEE/CVF International Conference on Computer Vision (ICCV), 2025

Harnessing Vision Foundation Models for High-Performance, Training-Free Open Vocabulary Segmentation

Yuheng Shi, Minjing Dong, Chang Xu

[Code]

IEEE/CVF International Conference on Computer Vision (ICCV), 2025

Multi-Scale VMamba: Hierarchy in Hierarchy Visual State Space Model

Yuheng Shi, Minjing Dong, Chang Xu

[Code]

Conference on Neural Information Processing Systems (NeurIPS), 2024

YOLOV: Making Still Image Object Detectors Great at Video Object Detection

Yuheng Shi, Naiyan Wang, Xiaojie Guo

[Code]

AAAI Conference on Artificial Intelligence (AAAI), 2023

📄 Preprints

Q-Zoom: Query-Aware Adaptive Perception for Efficient Multimodal Large Language Models

Yuheng Shi, Xiaohuan Pei, Linfeng Wen, Minjing Dong, Chang Xu

[Code]

arXiv:2604.06912, 2026

🎖 Honors and Awards

• 2025, Faculty of Engineering Research Support Scholarship, University of Sydney.
• 2023, 2nd Place, ICCV 2023 VCL Challenge: Robust Raw Object Detection.
• 2021, Outstanding Graduate, Central South University.
• 2018 – 2020, Second Class Scholarship, Central South University.

📖 Educations

• 2025.09 – Now, Ph.D. in Computer Science, University of Sydney. Supervisor: A/Prof. Chang Xu.
• 2021.09 – 2024.06, M.Sc. in Computer Science and Technology, Tianjin University. Supervisor: Prof. Xiaojie Guo.
• 2017.09 – 2021.06, B.Eng. in Intelligence Science and Technology, Central South University.

💼 Work Experience

• 2024 – 2025, Research Assistant, City University of Hong Kong. Supervised by A/Prof. Minjing Dong and A/Prof. Chang Xu.
• 2022 – 2023, Research Intern, TuSimple. Supervised by Dr. Naiyan Wang and Zehao Huang.

🛎 Academic Services

• Conference Reviewer: NeurIPS, ICLR, ICML, ICCV, ECCV.
• Journal Reviewer: T-PAMI, TMLR, Neurocomputing.