When it comes to facial recognition, age is beginning to matter a lot more as an estimated factor, particularly in the commercial and government sectors. Age-based access control, age verification, identification in missing persons, crime and disaster situations, and certainly age-based marketing are all facial recognition applications that modern technology is now able to provide solutions for.
The National Institute for Standards and Technology, otherwise known as NIST, has recently published test results on the performance of the automatic age estimation algorithms that make the above applications possible. Coming out on top with the most accurate age estimation for all age groups was Cognitec Systems, whose algorithm shows particularly impressive performance on youth and senior faces.
The test was conducted on algorithms submitted by six participants (five companied and a university). The nine algorithms that were submitted for testing had to identify seven million images and estimate the age range for each. The full test report can be found for perusal on NIST’s official website.
Cognitec’s algorithms are found working in the company’s FaceVACS-VideoScan product, which can determine the age ranges of people on a video stream. The major application for this (as any marketer reading this has surely already guessed) is in real-time advertising targeted at changing demographics.
With FaceVACS-VideoScan, a crowd can be analyzed by age range and gender so that an ad display can change to an optimized advertisement. It can also be an excellent way for retailers to measure the demographics of their clients.
In terms of access control, obvious applications for this sort of technology can be found in industries that cater exclusively to the age of majority (casinos, bars, nightclubs etc.) Facial recognition has already been deployed in some of these situations to provide access control for VIPs and members, adding age into the equation is an easy jump.
April 7, 2014 – by Peter B. Counter