What does your face say about you?
New technology capable of predicting a person’s age, gender and ethnicity from an image of their face, could have a profound impact on the way we work, live and play.
That’s the belief of the American software company, Percipo Inc, which is working on commercial implementation of new face description technology developed by the University of York’s Department of Electronics.
We are looking towards a broad range of applications including in the retail and security sectors
Professor John Robinson
Founded in 2008 with a mission to make machines more human, Percipo has worked with the York team to develop its own face description technology and has built three iPhone apps to demonstrate its capabilities - PhotoAge, PhotoGenic and ChickOrDude. The apps are available for download from the iTunes App store and allow people to browse their iPhone or Facebook photos estimating the age, attractiveness and masculinity/femininity of faces.
The new face description technology, which allows computers to analyse faces in a similar way to humans, is built on methods developed over the last six years by Professor John Robinson and colleagues in the Department of Electronics working in the Visual Systems Laboratory. Not only can the technology produce accurate predictions of age and gender from either a still or live video, but it can even identify and track expressions such as a smile or frown.
Professor Robinson, Head of the Department of Electronics, said: “We are excited to be working with Percipo to take the technology in directions we hadn’t imagined. The collaboration has started with recreational phone apps, and now we are looking towards a broad range of applications including in the retail and security sectors.”
Mehran Farimani, Percipo’s founder and Chief Executive Officer, believes that putting this technology in the hands of consumers is paving the way to its wide adoption within many products and markets. Ultimately his company believes that machines that can perceive will lead to applications that will have a profound impact on the way we work, live and play.
He said: “With over 700,000 installs worldwide and 12 million photos processed so far, we are introducing to the masses what machines are capable of doing today. We are actively working with partners to apply this technology to a number of markets, from intelligent cameras for consumers to automatic demographics measurement systems for enterprises.”
The technology works out how numerical values applied to facial features will vary based on a mathematical model (conditional density estimation), and classifier boosting which systematically combines very approximate estimates to produce accurate predictions.
Development work is taking place in retail demographics where the technology will allow accurate estimation of customer demographic data. Companies in the UK currently spend more than £100 million per year gathering data on their customers’ in-store activity, often using a manual approach to convert this into demographic statistics. This technology has the potential to automate this process.
The retail project is the result of three years work in the Department, part-funded by the Yorkshire Concept Fund. Part of the ongoing research is supported by the Face Quiz. This collects information about how we perceive age, with the results helping to refine the estimators used within this research.
About the researcher
Professor John Robinson is Head of the Department of Electronics and Leader of the Visual Systems Lab, part of the Intelligent Systems Research Group. The Group concentrates on the development of adaptive algorithms for robust video processing. Professor Robinson’s biography is available at www.elec.york.ac.uk/staff/jar11.html
Further information
- More on the University of York’s Department of Electronics at www.york.ac.uk/electronics
- More on Percipio Inc at www.percipo.com