Demonstrating Eye Gaze Tracking using Computer Vision -

30 May 2017

Demonstrating Eye Gaze Tracking using Computer Vision

Author: Angela Rosa, Embedded Software Developer

Eye gaze tracking is a powerful computer vision capability that allows users to provide input and guidance to computer systems user interfaces by using merely eye movement. The topic of eye gaze control has been documented thoroughly [1], [2]. This blog post provides an overview of the sample Gaze Tracking application that is available on Intrinsyc’s Open-Q™ 820 Development Kit. This capability was demonstrated in our video posted last month:

Supporting this technology on our development kits gives developers a starting point to develop their own unique eye-tracking computer vision solutions.

Angela

The sample Gaze Tracking application running on our Open-Q 820™ Development Kit demonstrates gaze tracking and facial recognition performing remarkably well. As seen in the video the three dots on the screen indicate the detected positions of the person’s pupils and mouth. A fourth dot on the screen provides a visualization of the person’s current gaze point.

Furthermore, a table of measured real-time values can be seen on the screen to display the finer measurements taken by the application. In addition to gaze tracking, the Gaze Tracking application also processes other facial data and displays values in real-time relating to:

  • Smile detection
  • Orientation of the head
    • Including degrees of roll, yaw, and pitch
  • Blink values for each eye

The Gaze Tracking application is based on a sample application provided in Qualcomm’s Snapdragon computer vision SDK modified to highlight the gaze tracking technology and to provide a simplified display of the facial data processing points. The SDK includes a simple API for retrieving captured values from the camera subsystem that enables feature-rich computer vision processing scenarios [3]. In the Gaze Tracking application, instantaneous facial data values are retrieved at regular intervals to update the camera preview with real-time data collected for each detected face in the camera’s view.

All values collected using the facial data processing capabilities of the Snapdragon SDK are given on a scale, in degrees or point values, which allows for finer details of the face’s state to be measured. As a result of this a developer can create applications that respond to virtually any number of facial gestures.

For more information on the Gaze Tracking application and the Qualcomm™ SDK, please contact Intrinsyc. We offer a range of development kits where Qualcomm’s computer vision technology can be utilized. We have experience with OpenCV and Qualcomm’s FastCV™ and can help you with your hardware, software or product development.

References
[1] Chen, B. X. (2013, March 12). Eye-Tracking Tech Will Be Open to iPhones and Other Devices. Retrieved May 30, 2017, from https://bits.blogs.nytimes.com/2013/03/12/eye-tracking-tech-will-be-open-to-iphones-others/?_r=0
[2] Burgett, G. (2016, June 24). MIT researchers develop eye-tracking program for your smartphone. Retrieved May 30, 2017, from https://www.digitaltrends.com/mobile/researchers-mit-georgia-eye-tracking-program-app/
[3] Lukas, S. (2013, November 7). Using Facial Processing with the Snapdragon SDK for Android. Retrieved May 30, 2017, from

Author: Angela Rosa, Embedded Software Developer


Comments are closed.

Job Id*

Your Name*

Your Email*

How did you find us?*

Location*

Resume