Computer Vision Research

 

In 2007-2008, I worked as a user experience researcher in Nokia Research Center. The researchers I worked with developed applications that used cutting-edge computer vision technology in Nokia phones. I performed field studies on two different applications that used the same underlying technology.

 

 

While at Nokia Research Center, I helped publish two research papers in leading conferences.  The first paper, published in CHI (the leading conference in human-computer interaction) 2008, studied the use of a Nokia phone as a mobile tour guide. In these walking tours, people would capture images using the phone’s camera, then those images would be matched against a database of points of interest along the walking path. If a match came up, a short web page containing historical information about the object would be available to read.

The results provided interesting insights into how people might comprehend and use similar computer vision applications. In particular, we learned that piggybacking this new type of interaction on top of existing photo-capturing practices resulted in a better tourism experience. Working with my colleagues, I developed the research protocol, performed the study, analyzed the results, and drafted the research paper.

You can read the paper here.

 

The second paper, published in Mobile and Ubiquitous Media (MUM) 2008, studied the use of a Nokia phone as a navigational system. In addition to the traditional map-based wayfinding, we included a landmark-based series of pictures with overlaid arrows to help guide participants. The focus of this paper was the technical details of the underlying technology that automatically generated the landmarks and representative images of them, but the user study that I designed, performed, and analyzed was used to verify the functionality of the technology and inform future areas to explore.

For example, participants suggested adding turn-by-turn directions to the application, along with translating the directional arrows into a more realistic three-dimensional perspective.

You can read the paper here.