John Tsotsos

Canada Research Chair in Computational Vision

Tier 1 - 2017-11-01
Renewed: 2017-10-01
York University
Natural Sciences and Engineering Research Council

416-736-2100 ext. 70135

Research involves

Computational modeling of visual processes and visuo-cognitive behaviors so that the model has predictive power for human vision and applied relevance for practical robots.

Research relevance

This research will improve understanding of visual information processing in humans, machines and robotics. Applications include companion robots for the elderly and in autonomous cars.

Why is our sense of sight so effortless?

Normally-sighted people take vision for granted. But how exactly does vision happen? This question has occupied philosophers and scientists since ancient times. Computer scientists joined this quest in the early 1960s. But, rather than pursuing the question solely through introspection or human experiments, computer scientists attempted to build machines—cameras connected to computers—with the goal of making them see the way humans do.

Success has come slowly, but with rapid advances in technology "computer vision" applications are now common in cellphones, security systems, self-driving cars, and more. So, does this mean we now understand how vision works? No, far from it.

While scientists have come a long way by developing these specific applications, these devices don’t have the enormous breadth of function that human vision has that allows us to perform a wide range of actions—from reading a book to finding lost keys to playing a game.

As Canada Research Chair in Computational Vision, John Tsotsos' research attempts to develop theories on how human vision is possible and how for humans, it feels so effortless.

The impact of his research can be far-reaching. Current applications that Tsotsos is studying try to incorporate elements of human vision. For example, he is looking at how to make a companion robot for the disabled or elderly have broader functionality to be able to reason about what it sees, the way a human caregiver can. And, how an autonomous car can better anticipate the actions of people around it to increase safety.