Recognizing relatedness in the face – could a super-recognizer perform better than software?

I believe that I probably could out-perform both of the interesting kinship recognition computer programs that are discussed in this most interesting recent article in New Scientist magazine, but there is still a lot to be learned from the researchers’ efforts to develop software that can detect whether two people in photographs are related. The first program described in this article had a success rate of 68% in detecting parent-child matches, the second program a 71% success rate, and people (presumably with average ability in face recognition) had a success rate of 67%. I’d love to see a similar study done with super-recognizers compared to normal human controls or pitted against a computer program. My money would be on nature’s best, rather than the latest technology.

It’s interesting to note that the first program described uses a method which I think is very different to natural face recognition, analysing fine details of the picture, while human face recognition is thought to be a process of identifying an overall pattern rather than looking at the picture piece-by-piece. The focus on details of this program appears to give the program a fair degree of robustness and flexibility in dealing with variations in the appearance of faces in photos. As researchers have found in another study (published online in the journal Cognition) , there can be considerable variability in the appearance of the same face in different photographs, and while people who are familiar with a face are able to identify the same face in different photos, people who were unfamiliar with that face did not have whatever it takes to be able to overcome variations among different photos of the same person to identify the photos as being of the same person. Clearly there is something in the memory of a person who is familiar with a face which gives their ability to recognize that face a great robustness and flexibility. I wonder what it is?

A quote from the New Scientist article: “Lu reckons that improved algorithms could be used to help determine kinship when DNA testing isn’t an option. “It can also help refugees find dispersed family members,” he suggests.” I believe that with my well above average ability in face recognition I would probably perform well in these types of tasks. I would love to work in a job or a business in which I was identifying faces or picking out related people from photographs or identifying people who have a genetic similarity (I can be contacted through leaving a comment on a blog post). Face recognition is an ability with applications that go way beyond personal socializing. Superior face recognition could be useful in many important areas of work, including law enforcement, private detective work, social work (working with adopted people or displaced families) and medicine (identifying genetic syndromes), and it appears that the current state of technological development of face photograph recognition technology is at a pretty basic level, only marginally better than the ability of the average person. In contrast, super-recognizers have a face recognition ability that far exceeds that of normal people, and I have good reason to believe that I could be a superrecognizer. I feel quite confident about my ability to detect kinship or genetic similarity from looking at people’s faces because I believe that is probably what I was involuntarily doing when I experienced The Strange Phenomenon, which I have described in the first post in this blog. Last month the British Sunday Times newspaper reported that London’s Metropolitan Police force have an elite squad of super-recognizers who have proven to be much more useful than face recogition technology. “The Met” are actively searching for more superrecognizers within their ranks to help with the huge task of identifying faces in many hours of CCTV images of the English riots. It seems odd to me that so much research is being done on creating and improving technology in face recognition when we have only just started to understand the naturally-occuring human ability in face recognition that has always existed.

Facial recognition software spots family resemblance
7 December 2011 by Kate McAlpine
Magazine issue 2842
New Scientist

Rob Jenkins, David White, Xandra Van Montfort, A. Mike Burton Variability in photos of the same face. Cognition. Available online 3 September 2011.

A Most Peculiar Experience

Face photographs unsuitable as proof of identity due to within-person variability?

Post a comment or leave a trackback: Trackback URL.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: