VIU LogoVision & Image Understanding

Perception, covert attention and eye movements during visual search

mafc mammofishermanVen
eyes nat imageeyetracker

Top left image: Multiple forced choice study to investigate how radiologists detect tumors in x-ray mammograms; Top right: Fisherman scrutinzing the ocean surface searching for school of sardines in Margarita island Venezuela, Bottom left: Eye movements recorded with an infra-red video based eye tracker; Bottom right: Video based eye tracker in the VIU lab

Whether looking for your house key, a friend in a crowd or your car in the parking lot, visual search is an important task in your daily life. In addition, visual search was important for survival of ancient humans foraging for food and is life-critical for specialized tasks of modern man such as looking for tumors in a medical image and/or potentially dangerous objects in a baggage x-ray inspection. Visual search involves moving our eyes to examine different parts of the scene with the high resolution fovea and shifting our attention to regions of interest. Our laboratory focuses on trying to understand the brain mechanisms by which humans deploy attention and eye movements and identify the strategies and neural computations that the human brain uses to attempt to optimize visual search. Our premise is that computational models with eye tracking technology can be used as tools to rigourously test theories and bridge across the various disciplines studying search, attention and eye movements (psychology, cognitive neuroscience and neurophsyiology). In addition, we study search in real world tasks such as search for tumors in medical images and fishermen looking for school of sardines and/or tuna on the ocean suface.

Our research has concentrated on using Bayesian Ideal observer models and other signal detection theory based models to predict human performance and comparing these models to computational implementations of other models in the literature including limited resources, serial attention, and guided search models.

In addition we have developed web-based implementations of many of the popular visual search models and some of our own models for visual search accuracy. Cyberfits (click on logo below).


More detail about individual projects in three not mutually exclusive areas
(some publications are listed in various sections) and a recent review article


Visual search

Covert Attention

Eye movements during search