Research Interests

Computer Vision, Semi-supervised Learning, Robust Estimation, Geometry in Machine Learning, AI for Conservation, Sustainability and Society

Research

My research interests are in the design of solutions to computer vision problems, which often include machine learning and statistical learning approaches. I am currently working on problems in Visual Wildlife Monitoring, Autonomous Driving and Road Safety and Sustainable Agriculture. Most of these problems require decision making based on lower-level learning tasks like object detection, segmentation, tracking, superresolution and domain adaptation.

Our group's focus is on designing learning algorithms that can work with limited amounts of labeled data by leveraging constraints from domain knowledge and geometry. We frequently use tools from Riemannian geometry, robust and nonparametric statistics, signal processing and optimization and combine them with modern machine learning techniques like deep neural networks.