Will a facial expression recognizer help autistic kids?
Computer scientists in Singapore are developing a system that locates the face using derivative-based filtering and then calculates which emotions are being expressed.
Computer scientists at Nanyang Technological University in Singapore are developing a system that could help autistic children identify the emotions of those around them by first locating the edges of faces, then finding crucial fiducial points to extract and process features, and finally classifying those features into corresponding emotions.
Their paper, "Towards a Portable Intelligent Facial Expression Recognizer," is available online through the journal Intelligent Decision Technologies (Volume 3:3).
"Emotion is a state of feeling involving thoughts, physiological changes, and an outward expression," write the paper's authors, Teik-Toe Teoh, Yok-Yen Nguwi and Siu-Yeung Cho of the Centre for Computational Intelligence of the School of Computer Engineering of Nanyang Technological University. "In this paper, we propose a system that synergizes the use of derivative filtering and boosting classifier."
One theory related to autism, developed in the 1980s and known as empathizing-systemizing, suggests that autistic individuals can systemize (develop rules of operation to handle events within the brain), but not always empathize (develop rules of operation to handle events generated by external agents, i.e. another person's face).
Whether or not the portable device will effectively, efficiently, and discreetly translate what is going on in the faces around autistic children (or adults, or perhaps anyone who struggles with interpreting facial expressions) remains to be seen. So does whether it helps with empathy, or whether a significant number of people even want help in this area. We'll report back if and when this hits the market.
In the meantime, the amateur poker player within me--and I cannot be the only one curious about this--would love to see which emotions this device would find at a high-stakes poker table full of the world's best poker faces. For purely scientific reasons, of course.