“Lie To Me” Automated: Computers Detect Micro-expressions and Lies

“LIE TO ME” AUTOMATED: COMPUTERS DETECT MICRO-EXPRESSIONS AND LIES

A leading computer vision research groups has succeeded to automate micro-expression detection, familiar from the TV series “Lie To Me”. Micro-expressions are very short involuntary facial expressions that reveal emotions people try to hide. They can be used for lie detection and are actively used by trained officials at US airports to detect suspicious behavior. For example, a terrorist trying to conceal a plan to commit suicide would very likely show a very short expression of intense anguish.

An interesting point about micro-expressions is that the human recognition accuracy is very low. Even highly trained human detectors are notoriously inaccurate. This makes an automatic computer detector very attractive.

“To the best of our knowledge, we are the first computer scientists to succeed in recognizing micro-expressions”, says Professor Matti Pietikäinen, leader of the Machine Vision Group at University of Oulu. “In addition, our system’s accuracy compares well with humans”.

Tomas Pfister, the University of Oxford scientist leading the project, says: ”I have been developing this idea for the past five years, ever since I read a book by psychologist Dr Paul Ekman. Together with my Oulu colleagues we made the final breakthrough earlier this year”.

Because micro-expressions are very rapid (1/3 to 1/25 seconds), current computer vision methods cannot reliably recognize them using a normal camera. The team solved the problem by introducing an interpolation method. They then used the Oulu group’s state-of-the-art Local Binary Pattern (LBP) method to characterize micro-expressions numerically.

A paper detailing the method was published in the 13th International Conference on Computer Vision this November. The new micro-expression database is available to the research community.

Images, videos and the full research paper are available on request.

The Oulu Machine Vision Group is world-renowned for its expertise in computer vision, particularly for its highly successful Local Binary Pattern (LBP) methodology. Tomas Pfister is with the world-renowned Visual Geometry Group at Department of Engineering Science, University of Oxford.

Contact Information:
BA (Cantab) Tomas Pfister, Department of Engineering Science, University of Oxford, tp@NOSPAM-robots.ox.ac.uk
Prof Matti Pietikäinen, University of Oulu, tel. +358 40 589 0450, matti.pietikainen@ee.oulu.fi
Hannakaisa Aikio, University of Oulu, tel. +358 400 142 293, hannakaisa.aikio@ee.oulu.fi
Department of Computer Science and Engineering, PO Box 4500, FI-90014 University of Oulu, Finland
Website: http://www.cse.oulu.fi/MVG

This release was published on openPR.