Eye Localization

Submitted 8 years ago

We have developed a functionality for mobile Android devices which accurately localizes the eyes of the user, using the frontal camera of the device. The eye localization functionality works in real time and detects the eye centers robustly and accurately, even in images of lower resolution, regardless of the illumination conditions (given that the eyes are visible). A demo of the eye localization functionality is published in Google play https://play.google.com/store/apps/details?id=org.opencv.eyelocvasko

The eye localization algorithm is based on extensive research performed in the frames of my PhD thesis and it has demonstrated superior performance compared to other similar algorithms. Its description and comparison with other state-of-the-art eye localization algorithm is published in international journals and conferences ( http://www.sciencedirect.com/science/article/pii/S026288561500013X and http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6495155&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D6495155).


APPLICATION AREAS

The development of eye localization was initiated as part of a project to develop a robust platform based on low cost hardware, with the ambition to integrate its functionality in other applications such as:


• Smart scrolling
Similar to Samsung's Smart Scroll feature which currently relies on face recognition and tilt (http://mashable.com/2013/03/14/smart-scroll/), smart scrolling can be further optimized by sensing the change of the direction of the gaze.

• Spoof proof face unlock
The face unlock feature can be usually spoofed using a photo of the owner (http://www.geek.com/android/android-face-lock-feature-spoofed-by-photograph-1440953/). Face unlock can be rendered more reliable by introducing a 'liveness check': once a face is verified, the user is required to follow a randomly moving dot with his eyes.

• Estimation of gaze direction for driver assistance
By mounting the mobile phone on a vehicle's dashboard, a gaze estimation application could monitor and evaluate the driver's vigilance, by determining whether the driver's focus of attention is on the road.


An Android application that enables typing using the gaze direction, has also been implemented using the eye localization functionality. The "Eye Type" application participated in two international competitions and was distinguished from both with the following awards:
1. Computer Society Award (IEEE Mobile Application Development Contest 2015) http://ieeemadc.org/ieeemadc/
2. Most Innovative App prize (EESTEC International Android Competition 2015) https://competition.eestec.net/

A demo application can be found in this link https://www.youtube.com/watch?v=6KOoxkY7KBc