I am interested in developing novel computer vision and machine learning algorithms to solve visual analysis problems arising from diverse real-world applications. Majority of my current research can be categorized into three major thematic areas -
Data-efficient Deep Learning: zero-shot/few-shot knowledge distillation, continual learning, source-free/unsupervised domain adaptation, data-free adversarial defense, privacy-preserving ML etc.
Cross-modal Computer Vision: text-based image retrieval, sketch-guided localization, scene-graph generation, visual question answering etc.
Video Surveillance: robust and scalable person-reidentification, anomaly detection, human gait and pose analysis etc.
In the past, I have extensively worked on biological and medical image analysis problems (e.g., registration, cell segmentation, spatio-temporal cell tracking, cell and tissue resolution 3D reconstruction of confocal image stacks, analysis of clinical MR and CT images etc.). I am also interested in data-association problems on graphs and their applications to visual analytics tasks.