John Tsotsos


Canada Research Chair in Computational Vision

Tier 1 - 2003-10-01
York University
Natural Sciences and Engineering

416-736-2100 ext. 70135
tsotsos@cs.yorku.ca

Research involves


Using computer science, mathematics, humans, and robotics to further understand vision and visual attention in biological systems and computer vision systems.

Research relevance


The research promises to advance our understanding of the mechanisms behind human vision, to design better tools for medical diagnosis, and to create robotic devices to assist disabled children and adults

Computers That See


As the Canada Research Chair in Computational Vision, Dr. John Tsotsos is looking for ways to model human mechanisms of visual motion in machines. Integrating the fields of visual psychology, computer vision, robotics, and visual neuroscience, he is developing robotic wheelchairs for the disabled that can be controlled by vision. His research in active vision (computer vision systems equipped with cameras that move and attend to items of interest), motion recognition, and mobile robotics will benefit Canadian industry and the health sciences by advancing software and hardware design and application, as well as by developing better medical diagnostic tools and biomedical visual and motor devices for use by disabled children and adults.

Dr. Tsotsos has led many experimental efforts in active vision. As well, he designed and implemented the first computerized motion recognition system, showing that visual motion converted into digital image sequences can be automatically detected, quantified, and interpreted by a computer. This work was applied to cardiology, providing the first example of an automated system to assess the performance of the left ventricle from X-ray image sequences.

In addition, Dr. Tsotsos showed for the first time that methods of theoretical computer science could be applied to the analysis of biological vision, leading to computer processing architectures for explaining vision problems in the brain. He was the first to formally prove that tasks or specialized knowledge play a critical role in managing human perception. He developed the Selective Tuning Model (STM) for visual attention, which is widely considered the leading model for consolidating current understanding of the process of visual attention. And he applied this model in machine vision. The STM theory of visual attention, together with subsequent experimental evidence, has formed the basis for greater understanding of human perception in the fields of psychophysics and neurobiology.