[Home] [Research] [Students] [Teaching] [Contact]
[Research] [Prospective Students] [CPL]

Information for New Students and Post-Docs interested in working with me.

I am always looking for bright and very motivated students (graduate and undergraduate) and Post Doc to work with me on exciting research projects.

While there are a variety of backgrounds that are appropriate for working in my lab. As strong a background in mathematics as possible is good, as is the ability to code real systems.  Experience with machine learning is also helpful.

Here I outline some of the areas in which I would be very interesting in finding folks to collaborate with:

Aware Home

This is a long-term exploratory project aimed at understanding the impact of future technologies (that we investigate, build, deploy, and test) to the residential setting.  Possible applications being envisioned involve the support of the elderly residents to remain active and independent. Primary focus of my research effort here is building awareness into this space.  We are building various sensing (vision, audio, etc.) systems to track and recognize both short-term and long-term behaviors of residents. Current projects include:

• Tracking using multiple cameras
• Recognition of activities (what is a person doing?)
• Modeling interactions between people.

More Information on the Aware Home is available from its Web Page.

Human Activity Recognition

The work here is currently focused on two basic areas:

• Action anomaly  detection: Surveillance

See the CPL Action Recognition site for details.

• Probabilistic sequence recognition

Both within the context of the above mentioned Aware Home Project and in the context of everyday activities, one question is very important.  "What is a person doing?" "   I am very interested in exploring vision-based  methods for interpreting and recognizing human behaviors and in limited settings predicting what is going to happen.  For some recent work see Yifan Shi's work also on the CPL Action Recognition site.

Video-based Rendering and Animation

This project will extend our following previous work

• Multi-dimensional Texture Synthesis using Graph Cuts: A patch-based method for texture synthesis in multiple dimensions. An algorithm for finding optimal seams between input and output texture (for a given texture placement/offset) is developed. Our technique allows iterative refinement of already generated textures.

My group publishes primarily in Computer Vision, especially in CVPR and ICCV.  Occasionally publish in HCI and Machine Learning venues as well. Recently have begun extensive collaborations in publishing with other members of CPL - this is the most fun.

[Home] [Research] [Students] [Teaching] [Contact]
[Research] [Prospective Students] [CPL]

This page was last updated on 01/19/09 03:05:26 PM.