Information

Activities view
Coursework view
Resume (.pdf) view
Contact view

Interests

Yelp view
Last.fm view
The New Yorker view
xkcd view

Other

Credit: syndicateme.net OSWD more
Tiddlywiki based technical specification on sourceforge more

Activities

Intel First Year Research Experience (IFYRE)

In Spring 2008, I began working with Professor Martial Hebert on a project relating to event detection in crowded videos. I received a crash course on image representation, manipulation, and general computer vision essentials. I learned how to use Matlab, specifically the Image Processing Toolbox, to modify and test image segmentation algorithms. My project focused mainly on segmenting videos using watershed segmentation and agglomerative clustering segmentation and then running the results (called labels) through a shape-matching function. The shape-matching function compares a template video for a specific action with the labels to determine the location on each frame that best matches the template.

The shape-matching function outputted three values for each frame: the distance ( the number of pixels that the template did not overlap with the frame), and the x- and y- coordinates of the best match between the template and the frame. Designating a threshold of 5000 pixels to be considered as an identification (a distance of less than 5000 pixels would be considered an identification of the event), I manually determined looked at each frame and determined true-positives, false-positives, true-negatives, and false-negatives to construct an ROC space for both types of segmentation. The results indicated that both types of image segmentation did not yield a consistently good true-positive rate. However, much room remains for improvement, particularly fine-tuning the parameters of the segmentation functions to be better suited for the types of videos being segmented.

Visual materials will appear here eventually.


Student Technology Outreach Club (STOC) website

I have been a member of STOC since the spring of my freshman year. STOC is a service organization whose mission is to provide technology help to nonprofit organizations on campus and in the Pittsburgh community. The club's projects include tutoring adults and younger students on computer usage, computer recycling, and web page creation. We are open to any project ideas that are consistent with the mission of the club.

I have been most involved with teaching basic computer usage to elementary school students. Every other weekend in the spring semester, we visit a local community center to teach the students how to use a particular application (usually Microsoft Office applications) or general computing skills like Internet safety, web design, and computer maintenance.