Chenghang Lai / Personal Homepage

Biography

I am a lecturer at Wenzhou University and the director of LingVision Lab. I received my Ph.D. from the School of Computer Science, Fudan University (2021.9–2025.6). My research focuses on multimodal large language models, world models, and multi-agent systems, with applications in general video understanding, medical imaging, industrial inspection, and sports analytics. During my doctoral studies, I was guided by Academician Qionghai Dai from Tsinghua University and Professor Xiangyang Xue from Fudan University.

Email / GitHub

LingVision Lab admits a limited number of students each year. Outstanding master’s and undergraduate students who are interested in our research are welcome to apply to join the lab.

🔬 Research Interests

LingVision Lab focuses on multimodal foundation models that perceive, reason about, and act upon complex visual worlds. Our research directions span general video understanding as well as medical, industrial, and sports applications.

1. Multimodal Large Language Models: Design and Applications

Designing multimodal LLMs that ground language in visual and temporal signals, and transferring these models to high-stakes application domains with reliable understanding and localization.

General vision–language understanding

  • Video understanding and temporal reasoning
  • Video question answering
  • Visual grounding and temporal localization

Medical imaging and surgical intelligence

  • Whole-slide image (WSI) analysis
  • CT analysis and medical image segmentation
  • Surgical video understanding and skill assessment

Industrial perception

  • Intelligent visual inspection
  • Anomalous event monitoring

Sports analytics

  • Athlete action quality assessment

2. World Models

Learning predictive models of environmental dynamics to support temporal foresight, planning, and decision-making in multimodal settings.

3. Multi-Agent Systems

Coordinating multiple agents for collaborative perception, reasoning, and task execution in complex environments.