Just as we applied Haar cascades to detect and track the presence of faces in an image/video stream, we can do the same for eyes.
The same general process can be used for eye detection as face detection, but instead of applying the eye detector to the entire image, we can increase the accuracy of the eye detector by first detecting faces and then only applying the eye detection to the face ROI.
Any time you can break your computer vision pipeline into steps and sub-steps such as these, you will likely obtain higher accuracy.
Simple, accurate eye tracking
A question I often get asked is:
“How do I track pupils in a video?”
I personally have not tried to build a pupil tracking system before, but I know that other PyImageSearch readers recommend and have had good experiences with the following tutorial:
http://thume.ca/projects/2012/11/04/simple-accurate-eye-center-tracking-in-opencv/