updated 3/14/2011 2:10:34 PM ET 2011-03-14T18:10:34

Thanks to new software that can identify a person’s mood through their facial expressions, computers may someday be able to empathize with their users.

The software, which was developed by researchers at Binghamton University in New York, uses images taken from several cameras to build a three-dimensional image of the user’s face.

Features of this image are then fed into a mathematical model, which spits out one of six human emotions — anger, disgust, fear, joy, sadness and surprise.

Potential applications for this software include using a person's facial expressions to trigger a computer action, or monitoring a person’s emotions to identify his or her satisfaction with a product, said computer scientist and lead researcher Lijun Yin.

Another application of the software might be as a lie detector test. The software could also help build avatars to teach autistic children how to read human emotions, Yin added.

So far, the researchers have created a database of 2,500 facial expressions from 100 subjects. While this is a big step toward computers understanding a user's emotions, there's still a ways to go, the researchers say.

"Some psychologists argue [that] these six expressions may not be very representative," Lin said. "More and more, we find we need to extend [the] six expressions to other expressions because people in daily life have more than the six classic expressions."

In their studies, participants sat in front of six cameras and made a facial expression corresponding to one of the six emotions for the computer. The next step will be using the model to identify spontaneous emotions seen in everyday life.

Yin collaborated with Peter Gerhardstein, a psychology professor at Binghamton, to develop the 3-D facial expression model.

•    Look, Ma, No Hands! Steering a Car Via Mind Control
•    For the Paralyzed, Wearable Robots May Bring More Interaction
•    10 Profound Innovations Ahead

© 2012 TechNewsDaily


Discussion comments


Most active discussions

  1. votes comments
  2. votes comments
  3. votes comments
  4. votes comments