My research interests are in object recognition, natural scene understanding, and using machine learning techniques to solve robot perception problems. I investigate the detection and recognition of objects in indoor household environments. I collaborate closely with the Intel Science and Technology Center for Pervasive Computing.

RGB-D Object Dataset

A large dataset of 300 common household objects recorded using a Microsoft Kinect style 3D camera. Visit the dataset website for more information and to download.

RGB-D Object Recognition and Pose Estimation

In this project we address joint object category, instance, and pose recognition in the context of rapid advances of RGB-D cameras that combine both visual and 3D shape information. RGB-D cameras are the underlying sensing technology behind Microsoft Kinect. The focus of this project is on detection and classification of objects in indoor scenes, such as in domestic environments. See the RSE Lab project page for more information.

Embodied Object Recognition

While completing my undergrad at UBC, I worked with a team of graduate students on a robot system capable of performing the Lost & Found task. Specifically, given a textual list of objects, such as red bell pepper and DVD "Gladiator", the system autonomously searches the World Wide Web for images of these objects, uses the images as training data to construct image classifiers, and then explores a real-world environment to locate the objects.

The system won first place in the Semantic Robot Vision Challenge in all three years in which the contest has been held (2007 and 2008 robot league, 2009 software league). More information on the team website.