Beiyuan Zhang

Hello, my friends! Welcome to my world!

I am a first-year Phd student of Computer Science and Technology, Beijing Institute of Technology, under the joint supervision of Prof. Ying Fu and Prof. Junliang Xing at Qiyuan Laboratory.

I obtained my M.S. in School of Big Data & Software Engineering at Chongqing University in 2025, supervised by Prof. Chunlei Fu and B.Eng in School of Electronic and Information Engineering at Dalian University of Technology in 2022. And I worked as a research assistant at the NLPR, Institute of Automation, Chinese Academy of Sciences from 2023 to 2024, supervised by Prof. Zhenan Sun.

My research interests lie in the intersection of Computer Vision and Machine Learning. In the past, I was focused on video generation, and now I have started exploring the delights of multimodal generation.

news

Dec 21, 2024	One paper is accepted by ICASSP 2025!

selected publications

Follow-Your-MultiPose: Tuning-Free Multi-Character Text-to-Video Generation via Pose Guidance

Beiyuan Zhang, Yue Ma, Chunlei Fu, Xinyang Song, Zhenan Sun, and Ziqiang Li

In Proc. of International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2025

Abs Bib PDF

Text-editable and pose-controllable character video generation is a challenging but prevailing topic with practical applications. However, existing approaches mainly focus on single-object video generation with pose guidance, ignoring the realistic situation that multi-character appear concurrently in a scenario. To tackle this, we propose a novel multi-character video generation framework in a tuning-free manner, which is based on the separated text and pose guidance. Specifically, we first extract character masks from the pose sequence to identify the spatial position for each character, and then single prompts for each character are obtained with LLMs for precise text guidance. Moreover, the spatial-aligned cross attention and multi-branch control module are proposed to generate fine-grained controllable multi-character video. The visualized results of generating video demonstrate the precise controllability of our method for multi character generation. We also verify the generality of our method by applying it to various personalized T2I models. Moreover, the quantitative results show that our approach achieves superior performance compared with previous works.
@inproceedings{zhang2025follow, title = {Follow-Your-MultiPose: Tuning-Free Multi-Character Text-to-Video Generation via Pose Guidance}, author = {Zhang, Beiyuan and Ma, Yue and Fu, Chunlei and Song, Xinyang and Sun, Zhenan and Li, Ziqiang}, booktitle = {{Proc. of International Conference on Acoustics, Speech and Signal Processing (ICASSP)}}, pages = {1--5}, year = {2025}, organization = {IEEE} }