A fast method for detecting eye accessing cues used in Neuro-Linguistic Programming

In this paper a simple and efficient method is proposed for detecting eye positions which are used in Neuro-Linguistic Programming (NLP) as clues for the human internal representational system and mental activity. The algorithm takes as input a square crop of a face and locates the eyes using a fast and efficient color-based method. The gaze direction is then detected based on the relative position of the iris and the sclera, using vertical and horizontal integral projections inside the bounding box of each eye.