Researchers create new eye-tracking system based on crowdsourced data

Researchers create new eye-tracking system based on crowdsourced data

Researchers at the Massachusetts Institute of Technology (MIT) and the University of Georgia have developed new software which is capable of turning any smartphone into an eye-tracking device.

The new eye-tracking system developed by the researchers depends on crowdsourced data. The system will be described in detail by the researchers in a paper which they will present at the Computer Vision and Pattern Recognition conference scheduled to be held on June 28.

The eye-tracking system created by the researchers essentially underscores a simple iOS app which flashes a small dot somewhere on the screen of devices running Apple’s iOS mobile operating system. The dot attracts the device users’ attention, and is briefly replaced with either an ‘R’ or an ‘L,’ asking the users to swipe either the right or left side of the screen. The tap on the dot by the users prompts the device to record video and images of the interaction.

For testing the eye-tracking system, researchers used Amazon's Mechanical Turk crowdsourcing site to employ 800 mobile-device users. Thereafter, the researchers used image processing techniques to assess the exact spot on which their eyes were pointed when tapping on a certain dot.

The researchers used machine learning technology to sense the eye position from the 1,600 photos which the eye-tracking system captured for each user.

Health