Projects
Multimodal Perception and Learning for Human–AI Interaction

• Embedded Computer Vision
  - Modeling vehicle-pedestrian interactions and real-time risk assessment
  - Real-time computer vision systems and synthetic data generation for UAV platforms
  - Multimodal perception for assistive autonomy: integrating human state recognition and vision-based wheelchair navigation
  - Advancing egocentric vision for human–AI interaction (e.g. hand pose estimation, document understanding)

• Medical Imaging Applications
  - Surgical AI for intraoperative decision-making: video-language models for workflow understanding and ultrasound-based physiological state assessment
  - Self-supervised representation learning for classification and segmentation to support early diagnosis and progression monitoring of retinal disorders
  - Multimodal learning for integrated healthcare decision-making