Welcome
to my homepage at UBC. I am a PhD Candidate studying with [Dr. Jim Little] at the Laboratory for Computational Intelligence in the Department of Computer Science. My thesis focuses on 3D scene understanding based on data from sensors such as the Microsoft Kinect, or robots like Willow Garage's PR2. I believe that modeling variables such as occlusion, viewpoint, and parts structure is most naturally done in three dimensions. Also, a number of useful priors such as the expected object's size, and the expected layout of objects in various typical scenes can most usefully be applied in 3D. I have developed several techniques to encorporate all of this information into an object recognition system, and I am continuing to study these problems to complete my degree.
This page contains information about my research, both past and present. Thanks for your interest. Please dont hestitate to drop me a line if you're interested in similar problems!
Upcoming IROS 2011 Oral
A paper describing our 3D object model that includes reasoning about 2D appearance and 3D geometry will soon appear at the [IROS 2011] conference. Click below to find the paper details.
PaperDataset Released
To encourage fair evaluation of robot recognition methods, we have collected the UBC Visual Robot Survey (VRS). It contains imagery and structure for many realistic scenes, along with human annotations of the present obejcts.
UBC VRSBMVC 2011 Poster
We have studied partial object detectors and explicit occlusion reasoning in a paper that appears in BMVC 2011. This work is collaboration with [Dr. Christian Wojek] and [Dr. Bernt Schiele] at the Max-Planck Institute in Saarbruecken, Germany,
Paper