John Geralds reports for vnunet.com. Excerpt:
Two US research teams developing software that can recognise and analyse facial expressions have caught the attention of the US Central Intelligence Agency (CIA), as a potential tool to build a better lie detector.
Professor Terry Sejnowski, who leads one of the research teams in the Computational Neurobiology Laboratory at the Salk Institute for Biological Studies in California, calls the new facial technology an “emotion detector”.
“It could be used in conjunction with a polygraph or more casually, for example, a camera hidden in the corner of a room. It distinguishes finer gradations of emotional response – whether the person is truly happy, sad or angry.”
Professor Sejnowski’s work coincides with that of Professor Jeffrey Cohn of the University of Pittsburgh and Carnegie Mellon University.
Cohn’s work is based on the coding system known as the Facial Action Coding System (FACS) developed in the 1970s. It defines the movements of each of the 44 muscles in the human face, information used by experts to study frames of video images and “read” people’s expressions.
“Our Automated Face Analysis system studies wrinkles and furrows, as well as other features, to quantify subtle changes in facial motion, rather than focusing on prototypic expressions,” explains Professor Cohn.