Zihao Liu

I am currently a joint Ph.D. student at Beijing Zhongguancun Academy (ZGCA) and Harbin Institute of Technology, Shenzhen (HITsz), supervised by Prof. Dr. Zhuotao Tian.

• Github: https://github.com/Kamino666
• Blog: https://blog.lab-kamino.com/

Research Interests

AI for Games, Video Understanding, Video Anomaly Detection, Multi-modal Large Language Models.

News

• 2026.04.10: Our new preprint paper of efficient & streaming & open-world VAD is available on ArXiv.
• 2026.01.27: One paper accepted to ICLR 2026.

Education

×

Beijing Zhongguancun Academy & Harbin Institute of Technology, Shenzhen

2026 – Present

Ph.D. Student · Joint Training Program

Supervisor: Prof. Dr. Zhuotao Tian

Communication University of China

2023 – 2026

Signal and Information Processing, Master's Degree

Supervisor: Xiaoyu Wu
GPA: 3.8/4.0

Communication University of China

2019 – 2023

Digital Media Technology, Bachelor's Degree

GPA: 3.8/4.0

Academic Service

• Reviewer for IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI).

Publications

ESOM: Efficiently Understanding Streaming Video Anomalies with Open-world Dynamic Definitions

Zihao Liu, Xiaoyu Wu*, Wenna Li, Jianqin Wu, Linlin Yang.

ArXiv, under review, 2026

[ArXiv] [Code]

Rethinking Metrics and Benchmarks of Video Anomaly Detection

Zihao Liu, Xiaoyu Wu*, Wenna Li, Linlin Yang, Shengjin Wang.

ArXiv, under review, 2025

[ArXiv] [Code]

Language-guided Open-world Video Anomaly Detection under Weak Supervision

Zihao Liu, Xiaoyu Wu*, Jianqin Wu, Xuxu Wang, Linlin Yang.

ICLR 2026

[ArXiv] [Code]

Enhancing Video Anomaly Understanding via Multi-Task Instruction Tuning

Xuxu Wang, Xiaoyu Wu*, Zihao Liu

IEEE Signal Processing Letters (SCI-Q2), 2025

[Link]

LoCo-MAD: Long-Range Context-Enhanced Model Towards Plot-Centric Movie Audio Description

Jiayi Wang, Zihao Liu, Xiaoyu Wu*.

ACCV'24 (CCF-C), 2024

[PDF] [Code]

Adaptively Building a Video-language Model for Video Captioning and Retrieval without Massive Video Pretraining

Zihao Liu, Xiaoyu Wu*, Shengjin Wang, Jiayao Qian.

ACM MM'24 (CCF-A), 2024

[PDF] [supp] [Code]

Violent Video Recognition Based on Global-local Visual and Audio Contrastive Learning

Zihao Liu, Xiaoyu Wu*, Shengjin Wang, Yimeng Shang.

IEEE Signal Processing Letters (SCI-Q2), 2024

[Link]

Semantic Multimodal Violence Detection Based on Local-to-global Embedding

Yujiang Pu, Xiaoyu Wu*, Shengjin Wang, Yuming Huang, Zihao Liu, Chaonan Gu

Neurocomputing (SCI-Q1), 2022

[PDF]