A Carnegie Mellon research team (see my comment) presented their ‘facial recognition’ findings at Black Hat this week (I commented on this study prior to Black Hat). They used publicly available data, which included photos from Facebook profiles (they use Facebook as an example) of students and then used facial recognition technology in an attempt to identify the students as they looked into a Web Camera.
Using a database of 25,000 photos from Facebook profiles, 31 per cent of the students (after three facial comparisions) were correctly identified. The research team then conducted another test using 277,978 Facebook profiles and compared them against approximately 6,000 profiles from a popular dating website. Amazingly, approximately 10 per cent of the dating sites members were able to establish their ‘real identity’. These are quite impressive findings.
Think about this – it’s much harder to change your face than your name – Minority Report anyone? Mapping an individuals online profile picture (assuming they do not use an Avatar or other non-facial image as their profile – that said some people post other pictures online, so it doesn’t have to be a profile image), and using street video cameras with facial recognition technology could very well provide anyone (and I mean anyone) with your ‘real identity’ as well as geo-tagging and timeline data. It’s the ‘timeline’ data streaming element that really fascinates me and I believe that in the distant future this ‘timeline element’ will play a major role in monitoring/tracking and marketing behavioural analysis.
What is the most anonymous data you have on the Web? If you have a picture or profile of “you” then it will be your face. Bet you didn’t know that! 🙂
The research team at Carnegie Mellon provide further evidence that academic minds (and others) are thinking far and wide about how easy it is to monitor and track individuals. It’s only a matter of time before a company or government decides there is value in using this type of technology innovation.
Safe surfing folks!