Currently, I am a third-year PhD student in Computer Science at the Computer Vision Group in the School of Electronic Engineering and Computer Science at the Queen Mary University of London, under the supervision of Prof. Shaogang Gong.
Before joining QMUL, I earned my Bachelor’s degree at Communication University of China and my Master’s degree at University College London.
My current research interests revolve around Deep Learning and Computer Vision, with a particular focus on Multimodal Video Understanding and Self-supervised Learning.
🔥 News
- 2025.09: 🎉🎉 One paper is accepted to NeurIPS’25!
- 2025.04: 🎉🎉 Our INT is accepted to IJCAI’25!
- 2024.07: 🎉🎉 One paper is accepted to ECCV’24!
📝 Publications
Uncertainty-quantified Rollout Policy Adaptation for Unlabelled Cross-domain Video Temporal Grounding, NeurIPS 2025
Jian Hu, Zixu Cheng, Shaogang Gong, Isabel Guan, Jianye Hao, Jun Wang, Kun Shao
- Unlabelled cross-domain temporal grounding
- Uncertainty-quantified Rollout Policy Adaptation with uncertainty-weighted rewards
V-STaR: Benchmarking Video-LLMs on Video Spatio-Temporal Reasoning, Arxiv 2025
Zixu Cheng, Jian Hu, Ziquan Liu, Chenyang Si, Wei Li, Shaogang Gong
ArXiv | Code | Webpage | HF Dataset
- Benchmarking Video-LLMs on Video Spatio-Temporal Reasoning
- Highlight the weakness of comtemporary Video-LLMs
CoS: Chain-of-Shot Prompting for Long Video Understanding, Arxiv 2025
Jian Hu, Zixu Cheng, Chenyang Si, Wei Li, Shaogang Gong
- Long-video understanding by visual prompt learning
- Training-free mosaicing binary coding with pseudo-temporal grounding for long video understanding
INT: Instance-Specific Negative Mining for Task-Generic Promptable Segmentation, IJCAI 2025
Jian Hu, Zixu Cheng, Shaogang Gong
- Training-free test-time adaptation for task-generic promptable segmentation
- Progressive negative mining identifies hard-to-distinguish error categories
SHINE: Saliency-aware HIerarchical NEgative Ranking for Compositional Temporal Grounding, ECCV 2024
Zixu Cheng*, Yujiang Pu*, Shaogang Gong, Parisa Kordjamshidi, Yu Kong (*equal contribution)
- LLM-driven methods for hard negative generation
- Coarse-to-Fine Saliency Ranking for Compositional Temporal Grounding
📖 Educations
- 2023.09 - now, PhD, Queen Mary University of London (QMUL), London.
- 2021.09 - 2022.09, Postgraduate, University College London (UCL), London.
- 2016.09 - 2020.06, Undergraduate, Communication University of China (CUC), Beijing.