Department of Computer Science Technische Universitaet Muenchen
Home People Teaching Research Publications Info Sitemap

Project EyeFinder

Matthias Wimmer

Abstract

The goal of this project is to support systems that aim to gain knowledge about their environment, involved persons, or even human intension. First of all basic image patterns have to be grasped. So the human face and the included eyes are to be detected here. It has been made use of Intel's OpenCV library.

Tracking the eyes
Movie 1: Tracking the eyes

Approach

  • locating the face using the OpenCV's face-classifier (Viola and Jones approach)
  • estimating the rough position of the eyes (Image 1)
  • improoving the location of the eyes using the eye classifier


The Eye Classifier

The eye classifier has been implemented using OpenCV's boosted cascade of haar-like features (just like the face classifier). The eye classifier is very weak. Due to the preceding estimation of the eyes' position it results in some great accuracy.

estimating the eyes' position
Image 1: estimating the eyes' position
tracking the face and the eyes
Movie 2: tracking the face and the eyes


Download

Here, you can download the locator for the eyes and for further facial components:

You can also download some example videos in order to get some information of how accurate the emotion detector works:
Movie 1: Tracking the eyes
Movie 2: Tracking the face and the eyes


Go back to the homepage of the smart environments research group.

Computer Science IX | Technische Universität München