Smartphone Selfie Camera Can Now Track Your Eye Movements

The talented engineers at MIT and University of Georgia have developed a new technology that can turn any regular smartphone camera into an eye-tracking device. The system can accurately determine the exact spot you are looking at on the screen and can be used not only to develop newer user interfaces, but also to detect signals of mental illness or neurological diseases. While eye-tracking technology isn't particularly new; for the last four decades it's been primarily used for psychological experiments and marketing related research; often requiring high-cost hardware systems.

Aditya Khosla (MIT) along with Kyle Krafka (Georgia Tech) and others built the new technology using advanced machine learning techniques that employ large number of data sets to determine the pattern. The team was motivated by the thought of making the technology available to regular users without needing special hardware.


The team used crowdsourcing to gather the data sets required to improve the accuracy of the system. The system analysed information collected from over 1500 smartphone users to get the accuracy to 0.5 cm; which the team believes is enough for commercial deployment.

The researchers used Apple iOS mobile operating system for collecting the information. Specially engineered app would flash a small red dot on the user's mobile screen just to attract their attention. Once the user notices the dot, the dot would change into R or L, instructing the user to swipe left or right on the screen. Successful completion of the task would inform the system that it grabbed user's attention. During the entire process - the front facing camera would keep capturing user's photos (~1600) to determine the exact direction in which the user is looking. Amazon's Mechanical Turk was used to gather information and each of the user was paid a small fee upon successful completion of the task.

The applications of this technology are numerous. It can be used to develop better and more intuitive user interfaces, in marketing research and also in helping doctors diagnoses special type of illnesses or diseases. Read more about the research on the official source.

Source: Eye-tracking system uses ordinary cellphone camera | MIT News | Massachusetts Institute of Technology