My current research lies at the intersection of computer vision and deep learning, with the focus on designing state-of-the-art techniques. Specifically, I am most interested in tackling challenging computer vision problems, including few-shot leaning for various detection and segmentation scenarios (pertaining to images, 3-D point clouds, or text-guided), fine-grained and long-tailed object detection and classification, image and video-based anomaly detection, video action and activity detection, deepfake detection, etc.

We have developed systems for COVID-19 CXR screening, brain metastasis segmentation, sepsis, and its prognosis prediction, detecting endotracheal tube and its positioning appropriateness, automatic beam angle selection for pencil-beam scanning proton liver treatments. We also established a federated learning platform to seamlessly promote joint research efforts on a wide spectrum of clinical applications across different medical institutes.