Algorithm Helps Machines Recognize Hidden Facial Expressions
Source: Tyler Lee
Seeing someone with a huge smile or grin on their face is usually a good indicator of pleasure. Seeing someone frown could indicate displeasure, sadness, or anger. These are pretty obvious facial expressions, but there is also another school that explores “hidden” expressions, or microexpressions which are basically small ticks or changes in our expression that give away how we really feel.
There is training that teaches us how to potentially pick up on these microexpressions, but why go through that effort when a computer can do that for you? Computers for the most part can pick up facial expressions (think smile detection systems on your camera), but recently researchers have developed a system that can magnify facial expressions, and in turn be able to pick up on hidden facial expressions.
Such a system would be useful especially if you want to know if someone is lying. However according to its researchers, the system is not perfect yet and as it stands, it cannot be used to spot microexpressions simply because the algorithm magnifies all movements as well, not just facial expressions.
It also has to rely on a training database to understand what it is seeing, so for now it isn’t exactly an autonomous system. However it is an interesting development and while lie detection in law enforcement is an obvious use for such a system, it could also be applied to job interviews in the future, or maybe used in wearables like Google Glass, the HoloLens, and so on.
| }
|