Math in the News
This Algorithm Can Reveal What You're Feeling
March 24, 2008
Researchers from Madrid's Universidad Politécnica and from the Universidad Rey Juan Carlos have come up with an algorithm that, by processing 30 images per second, can identify facial expressions in real time. It categorizes them as anger, disgust, fear, happiness, sadness, or surprise.
The system analyzes expressions through several boxes, each attached to, or focusing on, a part of a one's face. The boxes continually monitor facial movements until the algorithm manages to actually categorize an expression. This is done via careful comparisons with expressions captured from hundreds of sample expressions (333 sequences) from what is called the Cohn-Kanade database.
The algorithm's success rate, based on the Cohn-Kanade database, is purportedly nearly 90%. This algorithm has potential applications in human-computer interfaces and possibly in helping to improve e-commerce customer service.
The researchers were Luis Baumela and Enrique Muñoz, of the DIA, and José Miguel Buenaposada, of the Universidad Rey Juan Carlos de Madrid’s College of Computer Systems Engineering. They published their results in "Pattern Analysis and Applications" (January 2008).
Source: Science Daily