Automated Facial Expression Recognition with Application to Homeland Security


Littlewort, Frank, Whitehill, Wu, Lianscsek, Movellan, Bartlett


Our automated facial expression recognition work has profound consequences on law enforcement and counter terrorism. Careful laboratory studies show that many of the clues to concealed emotion and deceit currently used in law enforcement training programs may be quite unreliable. Moreover, research by Mark Frank, along with Paul Ekman showed that more reliable cues exist in facial behavior (e.g. Frank & Ekman 2004). Extracting this information requires detailed analysis of facial expression. Real-time automated coding can non-obtrusively supplement the other information available to interviewers, screeners, and law enforcement agents by identifying subtle or conflicted expressions that may betray someone’s true emotional state. Automatic Expression coding will also enable more thorough investigation of the role of facial expression in deception. The machine perception laboratory has worked with agencies involved in US security since 1999. 


NRL grant, PHI: Automated facial expression recognition with application to homeland security. PI: Bartlett 4/1/05-3/31/09.