Zongye Zhang

Currently a third year Ph.D. student at the Intelligent Recognition and Image Processing (IRIP) Lab from Beihang University, advised by Prof. Qingjie Liu.

My research interests revolve broadly across computer vision, generative models (diffusion models, vision-language models, etc.), and human-centric tasks.

Email / Scholar / Github

Research

Most of my work focuses on human-centric understanding and generation. Recently, I am exploring the integration of human-centric tasks with large language models (LLMs) to build more capable and interactive multimodal systems.

	Towards Robust and Controllable Text-to-Motion via Masked Autoregressive Diffusion Zongye Zhang, Bohan Kong, Qingjie Liu, Yunhong Wang ACM Multimedia, 2025 paper Exploring generating human motions in continuous representations, which supports arbitrary motion key-frames as input for better user control and generalizability. To achieve generation on continuous space, this paper employs masked autoregressive diffusion for generation.
	A Survey on Data Synthesis and Augmentation for Large Language Models Ke Wang, Jiahui Zhu, Minjie Ren, Zeming Liu, Shiwei Li, Zongye Zhang, Chenkai Zhang, Xiaoyu Wu, Qiqi Zhan, Qingjie Liu, Yunhong Wang ArXiv, 2024 paper This paper reviews and summarizes data generation techniques throughout the lifecycle of LLMs, including data preparation, pre-training, fine-tuning, instruction-tuning, preference alignment, and applications.
	SkeletonX: Data-Efficient Skeleton-based Action Recognition via Cross-sample Feature Aggregation Zongye Zhang, Wenrui Cai, Qingjie Liu, Yunhong Wang IEEE Transaction on Multimedia (TMM), 2025 arXiv / code
	SkeletonMix: A Mixup-Based Data Augmentation Framework for Skeleton-Based Action Recognition Zongye Zhang, Huanyu Zhou, Qingjie Liu, Yunhong Wang IEEE ICASSP, 2025 paper By selecting samples for mixup, the proposed method reduces the number of samples needed for training for existing GCN-based skeleton action recognition models.

Academic Service

Reviewer: IEEE Transaction on Circuits and Systems for Video Technology

This website use source code from this repo.