Research Focus

Currently, there are mainly two directions. The first is to conduct research on computer vision-based detection and recognition methods for micro-expressions (subtle, involuntary facial expressions that occur within 0.5 seconds and are difficult to detect). The second is to utilize large language models and multimodal language models for information extraction and visual question-answering tasks. In the medium to long term, the goal is to integrate language models with visual methods to explore model generalization and knowledge transfer from the real image domain to abstract and artistically exaggerated domains (for example, being able to automatically generalize from understanding real human faces to cartoon or comic-style faces). In recent years, we have also actively engaged in interdisciplinary research in the medical field, exploring ways to apply intelligent technologies to diagnosis and care.