UT Arlington College of Engineering
UT Arlington

Faculty Profile
Vassilis Athitsos – Picturing Silence



Communicating with deaf individuals not only means dealing with a lack of sound, but also with the way sign language is composed. Unlike a spoken or written language, American Sign Language (ASL) is not a literal representation of each word, so a complete sentence can be presented very rapidly. This often makes it difficult for signing novices to grasp the quickly changing hand positions and gestures.

But a research team involving Computer Science & Engineering Assistant Professor Vassilis Athitsos is developing methods to aid the learning of ASL by both deaf and hearing individuals.

Dr. Athitsos earned a Ph.D. degree in Computer Science from Boston University. Following his graduation in 2006, he remained at BU as a postdoctoral research associate working with Drs. Stan Sclaroff and Carol Neidle of Boston University’s Computer Science Department. As a postdoc, Dr. Athitsos conducted research on efficient similarity-based retrieval, gesture recognition, shape modeling and detection, and medical image analysis.

After coming to UT Arlington in 2007, Dr. Athitsos continued to communicate with Drs. Sclaroff and Neidle, hoping to secure funding to continue their work in gesture recognition. The team was successful; the National Science Foundation gave them a three-year, $900,000 grant.

Using computer vision, data mining and machine learning applications, the team began compiling a visual dictionary that will eventually contain approximately 4,000 commonly-used signs. This will allow non-ASL users to more easily match gestures with meanings and also study visual patterns. The team’s early endeavors have resulted in international media coverage. And we do mean international – from Kazakhstan to the Philippines.

At this point in their project, the team has collected an impressive amount of data. Three video examples have been collected, each signed by a different native ASL signer, for almost every sign in the Gallaudet dictionary of American Sign Language. Dr. Athitsos and his collaborators are now developing new gesture recognition techniques to achieve energetic performance with this massive dataset. The team is on target to produce in the next 12 months a working prototype that ASL signers and students can use to look up the meaning of an unknown sign.