Science News

Wearable Boredom Detector

Researchers at the Massachusetts Insitute of Technology (MIT) Media Lab are developing a mobile boredom detector designed to clip onto a pair of glasses and warn the wearer if they are annoying, or even boring the pants off the people they meet. The device is the brain child of MIT's Rana El Kaliouby and she hopes it will make life easier for people with autism who have trouble interpreting the body language of others. It works by relaying images of the facial expressions made by people chatting to the wearer to a handheld computer which analyses the pictures and picks out how the individual is responding to the conversation. Specifically the software focuses on movements of the eyebrows, lips and nose and also tracks nods and shakes of the head, and head tilting, changes in which can be boredom or mood-giveaways. If it spots any of these, the handheld computer vibrates alerting the user to change tack, or draw the conversation to a close. The researchers trained the computer by showing it over 100 eight-second video clips of actors displaying specific emotions. Now, when presented with prevously unseen video clips, the system correctly predict peoples' emotions 90% of the time when analysing actors, and 64% of the time when looking at shots of ordinary people. They're now training their software with footage from movies and webcams, and working on a way to shrink the camera and the handheld computer to a comfortable size.

2nd Apr 2006