It’s novel insight into this problem that earned Michael Cormier, a PhD student in the David R. Cheriton School of Computer Science, the 2018 Murray Martin Prize for Best Research Paper by a Math Grad Student. His winning paper "Purely vision-based segmentation of web pages for assistive technology," co-authored with supervisor Dr. Robin Cohen, is published in a special issue on assistive computer vision and robotics in the journal Computer Vision an Image Understanding.
Grounded in computer vision, a subfield of artificial intelligence research, this work hinges on the clever insight that web pages can be considered as visual images. Cormier used this foundation to design a novel computer vision model that segments web pages both hierarchically and optimally by using a Bayesian approach to detect edges, and which also considers classification.
Vision-based methods are not sensitive to implementation language or complexity, meaning this innovative model has applications far and wide in accessible technologies. His foundational work creating this model offers rich information and novel insights that form a viable launch pad for future computer vision research to take off from.
A direct application of this work to the field of assistive technology is improving systems that produce alternative presentations of text. It enables users, especially those with cognitive and visual challenges, to interact with web pages through a series of augmented experiences. Practical functions like decluttering pages or zooming preferentially, are desirable for everyday use.
The $3,060 Murray Martin scholarship is made possible thanks to generous donation from Pitney Bowes Inc. in honour of Murray Martin, the retiring chair, president, CEO and director whose continued investment in research and development has ensured the company’s industry leadership.