BASIC IMPLEMENTATION

* improve detection (set parameters depending on filesize / dimension)

ADVANCED FEATURES

* some kind of learning mechanism? (train the existing classifier, HEAVY STUFF!!!)

* preview? (select image from imagesList, open up a new window and select the already detected eyes.)
  Idea: If too many false positives are detected, select only the correct ones by clicking on these
        image regions.
        Clicking once: Turn rectangle around the eye green  = ok
        Clicking twice: turn rectangle red                  = remove the detected eye, learn from that mistake

ALL-PURPOSE BATCH DETECTION / CORRECTION TOOL

* use different classifiers to identify any object
  Idea: batch coloring of objects. For example you shoot a series of photos with an object in it. You
        want a different color for this object. Train a classifier to identify the object and let the user
        select a color that is overlayed above the object. REALLLLY HEAVY STUFF :-)
