OMRON incorporates “smile measurement” into facial recognition software
October 11, 2007 Depending on which set of figures you read, the words we speak are said to account for less than 10% of the meaning that we convey – tone, body language and facial expression make up the balance – but is it possible to accurately quantify these non-verbal aspects of communication? Facial recognition specialist OMROM has in part attempted to do just that, unveiling what is essentially a “smile detector” – a piece of software capable of objectively measuring smiles and giving them a percentage rating. On the surface it sounds like a novel idea, but there are some interesting practical applications for this technology ranging from market research to consumer electronics, where for example, cameras could be equipped with the ability to ensure that everyone in frame is smiling before a photograph is taken.
Smile measurement is the latest feature in Omron’s OKAO Vision suite of facial recognition software under development since 1995. The system uses a 3D model fitting technique to analyze faces - even in blurred, partially obscured or angled images – and can verify identity, estimate age and gender and track pupil or eyelid movements instantaneously.
The company claim the OKAO is also an unbiased piece of software and is capable of measuring the facial features of all ethnicities. The software can be integrated on a chip for mobile devices and is able read multiple smiles simultaneously.
The biometric software suite could also find applications in identity theft prevention, building entry management, driver monitoring systems in cars and access control for age-restricted content. It’s also suggested that the it could be used by mental healthcare professionals as a means of tracking patients’ mental condition or in market research to gauge consumer responses to new products.
The new technology made its debut at CEATEC JAPAN 2007. For further info visit the OMRON site.
Please keep comments to less than 150 words. No abusive material or spam will be published.