profile picture

Hejia Chen

Ph.D student

About Me

I am a Ph.D candidate with Beihang University, supervised by Prof. Shuai Li. I am also a research intern with Kling Team, Kuaishou. My research interests ground on facial behavior understanding and generation, and currently I am working on conversational avatar. I served as reviewer of ICLR, CVPR, ECCV and ECAI.

News

  • [2026/09/27]: I reached my 1000th day as a research intern with Kling AI.

  • [2026/06/18]: One first-author paper was accepted by ECCV 2026.

Show all 9 news

Publications

MindFlow: Harmonizing Cognitive Semantics and Acoustic Dynamics for Facial Animation Generation in Dyadic Conversations

H. Chen, H. Zhang, X. He, et al.

We tackle rigid and emotionally hollow facial expressions in dyadic conversation animation by enabling continuous emotion reasoning from streaming audio and seamless facial motion generation across both speaking and listening states.

Teaser
ECCV'26
Cafe-Talk: Generating 3D Talking Face Animation with Multimodal Coarse- and Fine-grained Control

H. Chen, H. Zhang, S. Zhang, et al.

We tackle the challenge of generating 3D talking face animations with controllable expressions. We propose Cafe-Talk, which integrates coarse- and fine-grained multimodal controls to disentangle conditions, synchronize lip movements, enhance expressiveness, and enable precise control.

Teaser
ICLR'25
Other publications (2)

X. He, H. Zhang, H. Chen, et al.

We tackle the identity drift, lip-shape leakage, and fragility of mask-based inpainting in audio-driven visual dubbing. We propose X-Dub, which repurposes an inpainting model as a pseudo-paired data generator to bootstrap a mask-free editing dubber, with timestep-adaptive multi-phase learning and a new X-DubBench benchmark.

Teaser
ICML'26

S. Li, S. Zhuang, W. Song, X. Zhang, H. Chen, A. Hao

We address the problem of synthesizing seamless human motion from sequential texts with semantic coherence and natural transitions. It uses a framework with L2G semantic fusion and transition reasoning modules, along with a custom transition loss and a new STDM dataset.

Teaser
ICCV'23

Education

Ph.D in Computer Application Technology

2022 - present
Beihang University (BUAA), China

I am currently a Ph.D student with State Key Laboratory of Virtual Reality Systems and Technology, Beihang University, supervised by Prof. Shuai Li. My research interest was Micro-expression Recognition (2022, BSc thesis w/ Xidian) and now I am working on Audio-driven 3D Talking Head.

BSc in Computer Science and Technology

2018 - 2022
Xidian University, China

I received my bachelor degree from School of Computer Science and Technology, Xidian University in 2022. I was a student research intern at Center of Intelligence-Aided Teaching (CIAT lab) from 2020 to 2022, during which my research interests are Student Engagement Prediction (2020) and Head Pose Estimation (2021).