ACM TechNews

Facial Expression Recognition Software

Universidad Politecnica de Madrid (02/25/08)

Artificial intelligence researchers at the Universidad Politecnica de Madrid have developed an algorithm capable of recognizing facial expressions in real time and categorizing them as one of six prototype expressions. The algorithm is able to process a sequence of frontal images of mocking faces, up to 30 images per second, to recognize a face's expression. The software can be applied to video sequences in realistic situations. The system analyzes facial expressions through several boxes, with each box "attached" to a part of the user's face. The boxes monitor the person's facial movements until it determines what the facial expression is by comparing captured information to different people from the Cohn-Kanade database. The system has a success rate of 89 percent and can work in adverse conditions where ambient lighting, frontal facial movements, or camera displacements create major changes in facial appearance.

http://www.sciencedaily.com/releases/2008/02/080223125318.htm

---

© Copyright 2008 Information, Inc. This service may be reproduced for internal distribution.