Machine learning used to distinguish between male and female smiles
Image credit: Dreamstime
A machine learning algorithm capable of identifying gender using just the movement of a smile has been developed by researchers at the University of Bradford.
This is the first algorithm of its kind; the vast majority of facial recognition algorithms use static images of the human face to determine age, gender and other characteristics.
The Bradford team began by mapping 49 features of the face, mainly around the eyes, mouth and nose. These ‘landmarks’ were monitored and used to assess exactly how the face is contorted by underlying muscles as we smile. The researchers then compared the movement in women’s smiles with that of men’s smiles, and found that women’s smiles tend to be wider.
“Anecdotally, women are thought to be more expressive in how they smile, and our research has borne this out. Women definitely have broader smiles, expanding their mouth and lip area far more than men,” said Professor Hassan Ugail, who led the study.
Based on what they found, the researchers developed a simple machine-learning algorithm. This proved reasonably effective at assigning gender according to a smile, correctly guessing gender in 86 per cent of cases.
“We used a fairly simple machine classification for this research as we were just testing the concept, but more sophisticated [artificial intelligence] would improve the recognition rates,” said Ugail.
According to Ugail, their study is less about categorising smiles and more about enhancing machine learning capabilities, although their choice of subject has raised some interesting questions, such as how the algorithm may respond to the smile of a transgender person, or a somebody who has undergone plastic surgery.
“Because this system measures the underlying muscle movement of the face during a smile, we believe these dynamics will remain the same even if external physical features change, following surgery for example,” Ugail commented.
“This kind of facial recognition could become a next-generation biometric, as it’s not dependent on one feature, but on a dynamic that’s unique to an individual and would be very difficult to mimic or alter.”
A range of algorithms exist which are capable of differentiating between the men and women by comparing features from static images, although some researchers have raised issues with these facial recognition algorithms; notably, that they fail to perform effectively when assessing portraits of women and people with darker skin tones.