Predicting Emotion
Dr Neil Mac Parthaláin

My work focuses on trying to automatically predict the six basic human emotions – namely happiness, sadness, fear, disgust, surprise, and anger – from the expressions of human faces on video.

The automated system is successful in doing so, and compares with human observers whom it agrees with 88% of the time. The expressions for which the automated systems does not score well are expressions on which human observers also disagree. Automated systems make it more efficient to do tasks previously performed manually by people, mainly to gather facial expression information related to a specific context where participants have consented to share such information. Emotion recognition can be used in society for a large number of reasons. For example, instead of filling out a long survey about how you feel about different segments of an educational video or advertisement, you may give permission to have a camera watch your face and to note when expressions such as happiness, surprise, disgust, or anger are demonstrated.