Eye-tracking research wins award
- 17 March 2016
Congratulations to Erroll Wood, a graduate student at Caius, who has received an international award for his work teaching computers to track and interpret human eye movements.
Erroll won the Emerging Investigator Award at the ACM SIGGraph Symposium on Eye Tracking Research and Applications for his paper Learning an appearance-based gaze estimator from one million synthesised images. The award was made at the symposium, held this week in Charleston, South Carolina.
Erroll is a research student working for a PhD in the Cambridge University Computer Laboratory under the supervision of Professor Peter Robinson, a Fellow of Caius. He previously studied Computer Science as an undergraduate at Caius with Professor Robinson as his Director of Studies.
Computerised tracking of eye movements has been developing for the last 40 years, Erroll says. But, he adds, “It’s still very difficult to track where people are looking in unconstrained environments outside the lab.
“We’re working at the moment on trying to teach a computer to accurately determine where someone is looking, from harsh outdoor lighting conditions to dark indoor conditions.”
Deep learning systems – networks based on a simplified model of the human brain – are used to interpret and classify the eye movement data. Erroll and his colleagues used computer-generated graphics rather than photographic images to create a million perfectly labelled images of eyes. Over time, their machine could learn how to classify these images.
Professor Robinson explains: “When people interact with each other, the eyes tell us a lot about what they’re thinking about. So if we can monitor someone’s eye gaze while they’re interacting with a computer system, we can probably see what they’re focusing their attention on and that might give us some hints as to the parts of the interface they’re finding difficult to understand.”
Applications for the technology could include simulating the control room of a nuclear power plant to ensure people can see visual alarms, or enhancing video conferencing to create better eye contact.
For more information contact: Communications Officer Lucy Ward; email@example.com