Faculty Profile: Marc Pomplun Works on Cutting Edge Human Vision Research

Associate Professor of Computer Science Marc Pomplun has become known around the university community for his cutting edge human vision research. How he originally got into the field, though, was somewhat of an accident.

While working on his computer science degree at Bielefeld University in Germany, Pomplun was focused on a career in robotics. But one day, a psychology professor who had purchased an eye tracking machine from Toronto asked him for help programming it.

“There was no software and no way for him to do anything with it, so he asked the computer science department for help. I thought ‘Yeah, well that sounds interesting,’ and I wrote some programs for it. He was very enthusiastic about it and we just kept working together.”

After earning his PhD in computer science, Pomplun continued his vision research in Toronto, as a post-doc in the psychology department at the University of Toronto and then as a research scientist at the Centre for Vision Research at York University, before coming to UMass Boston in 2002.

In the 10 years he has been here, Pomplun has seen a drastic increase in the amount and quality of research being conducted.

“When I arrived, we had six PhD students and now we have around 26. It’s quite a change,” he said. “Now there is much more assistance [in developing research grants] and people are much more knowledgeable.”

As director of the Visual Attention Laboratory, Pomplun researches human vision and the mechanisms that drive it. Using eye tracking systems called the EyeLink-II and EyeLink-2K, Pomplun and his students follow subjects’ eyes as they complete tasks, such as identifying all the blue objects on the screen.

“The interesting thing about it is not whether they’re actually finding the target, but where they are trying to find it,” he explained. “For example, if you’re looking for a blue item, are you only looking for everything blue in the image? Let’s say you’re looking for something that looks like a blue square, are you looking at everything that looks like a square or everything that looks blue?”

They use the information gathered to better understand how humans process visual information and develop computer models to simulate those processes, comparing the results to determine whether the models behave like the actual human subjects. “Our visual capabilities are still much better than those of any artificial system these days when it comes to object recognition or surveillance,” Pomplun said, but if they can develop accurate models, they can help build better artificial systems for people with vision problems.

The lab also uses a robot with a stereo vision system that simulates human eye movements. The robot searches for simple items like color blocks while researchers track its eye movements to see if they are similar to those of the human subjects.

“We wanted to see if the processes that we assume to occur in these different brain areas are actually responsible for our visual behavior. Based on what we’re seeing so far, we seem to have a rough idea of what’s happening.”

Pomplun explained that research using the robot goes beyond human vision. During space exploration, for example, it can take several minutes for a remotely operated vehicle traversing the surface of Mars to send visual signals back to the people on Earth who are driving it, meaning that it cannot avoid immediate physical obstacles. However, if the robot had its own vision system, it could recognize obstacles and act autonomously.

The robot was developed by Pomplun’s former student Tyler Garaas, who is now a research scientist at Mitsubishi Electric Research Laboratories in Cambridge. Garaas is one of many students to find success in the field after working with Pomplun. Since 2007, three students from the Visual Attention Laboratory have received their PhDs and another four have received their master’s degrees. Of the six undergraduate students who have worked with him in the past five years, four were recipients of the UMass Boston Undergraduate Research Funding Award.

Pomplun currently has three PhD students working with him, as well as undergraduate students from his courses, and is hoping to hire a post-doc soon. “I have found some really good students here. If you look at the best students in the course, I think that they could study anywhere basically. They are really good,” he said.

In addition to developing and teaching computer vision courses, Pomplun is working on a textbook called Hands-on Computer Vision, which will be published by World Scientific Publishers in 2013. Also an adjunct associate professor at Boston University, Pomplun said that the close proximity to other research universities and institutions is a benefit of conducting research at UMass Boston.

“People have been very supportive here,” he added. “Everyone has supported me from the beginning.”