Communications of the ACM
When Algorithms Grow Accustomed to Your Face
Some computer software can read subtle facial cues to determine a person's emotional state.
Credit: Koren Shadmi/The New York Times
Computer software currently exists that can read subtle, millisecond-long facial cues of a person's emotions via frame-by-frame video analysis.
For example, Affectiva developed a program from a database of about 1.5 billion recorded emotional reactions, and the software will be offered to mobile software developers beginning in mid-January.
Technology forecaster Paul Saffo says face-reading software may eventually be integrated with complementary emotion-recognition programs such as software that performs voice analysis. Advantages of such technology include smoother human-machine interaction, but some are concerned about the ramifications for surveillance and privacy infringement.
Meanwhile, Arizona State University professor Winslow Burleson envisions apps responsive to facial cues being broadly employed in medicine, gaming, education, and advertising. "Once we can package this facial analysis in small devices and connect to the cloud, we can provide just-in-time information that will help individuals, moment to moment throughout their lives," Burleson says.
Among those who he thinks could benefit from such technologies are autistic people and others who have difficulty reading facial expressions. For example, they could wear Web-linked goggles with cameras, receiving clues about the reactions of people with whom they are talking through an earpiece as the algorithm translates facial expressions.
From The New York Times
View Full Article
Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA
No entries found