Computer system learns to spot fake smiles
In the near future, it may be a lot more difficult to make computers believe that you're happy with something when you're actually not. That's because scientists at Britain's University of Bradford have developed software that can reportedly detect phoney smiles.
Led by Prof. Hassan Ugail, the researchers started with two pre-existing datasets – one of these (called MUG) consisted of videos of people who were smiling truly and spontaneously, while the other (called CK+) contained clips of people doing "posed" smiles.
Artificial intelligence-based algorithms were used to map the faces of the individuals in these videos, measuring the sequence in – and extent to which – their mouth, cheeks and eyes moved when smiling. While there were differences between the two groups in all three areas, the largest one involved the eyes. People who were genuinely smiling moved the muscles around their eyes at least 10 percent more than the fake-smilers.
Having learned from that assessment, the resulting software is now capable of assessing videos of smiling people, accurately determining whether or not their smile is the real thing. It is hoped that the technology may eventually find use in applications such as improved biometric security systems, better human/computer interfaces, and psychological studies.
"We use two main sets of muscles when we smile – the zygomaticus major, which is responsible for the curling upwards of the mouth, and the orbicularis oculi, which causes crinkling around our eyes," says Ugail. "In fake smiles it is often only the mouth muscles which move but, as humans, we often don't spot the lack of movement around the eyes. The computer software can spot this much more reliably."
The research is described in a paper that was recently published in the journal Advanced Engineering Informatics.
Source: University of Bradford