Perfect Harmony by Spyglass

In a slight panic, I noticed at the beginning of July that my passport had just 18 days left before it ran out.  I went to the government Passport Office website and to my delight, I found that renewal on-line was possible.  I began filling in the form.  To my delight, the form was really simple and user friendly and took just a few minutes to complete.  Great.  The final requirement was to add a digital photograph.  Not wishing to delay proceedings by getting the wrong picture, I scooted over to a professional photo studio in nearby Dorchester and £15 later I had the necessary e-photo in my in box, attached it to the application form and pressed send. Simples.

Then came the wobble. ‘Photo rejected’ came an error message telling me that the photo did not meet the parameters required by the Passport Office.  Having paid £15 and had it taken in a professional studio, I was just a tad annoyed.  During a grumpy phone call, the photographer confirmed that he did such pictures everyday and that it did indeed meet the requirement.

I rang the passport office to appeal – really helpful and efficient service.  My contact there confirmed the photo was in fact perfect – it was the electronic recognition system that had rejected it.  A common occurrence apparently. 


What if this error had been a wanted criminal, a murderer or a known extremist on a watch list?  Such errors by machines conducting recognition duties could have the most serious implications for safety and national security.  Electronic facial recognition systems are getting a bad press at the moment.  High error rates have led to some police services stopping their use during events, for example the Notting Hill Carnival.  This experience applies to police forces in the USA as well.

Facial recognition is an innovative technology and probably it will improve its performance as time goes on.  But at the moment, confidence is low and the huge investment in such systems has to be questioned until all of the bugs are ironed out.  There are also issues over the lack of regulation and approved operating procedures for such autonomous systems.

Whilst this debate is continuing, the use of Super Recognisers working alongside facial recognition systems must be a pre-requisite if public confidence is to be maintained.  Having a human in the loop could save me £15.  More importantly, it could save a wrongful arrest or a wanted person getting into the country illegally.  The concept of the 2 working in harmony is well described by academics including Dr Anna Bobak of the University of Stirling who said:

"The data…shows that combining the latest algorithm and a super-recogniser…leads to near-perfect performance"

Or, as a recent paper on Psychological and Cognitive Sciences dated 30 April 2018, published in the Proceedings of the National Academy of Sciences of the United States of America put it:

Therefore, collaboration among humans and between humans and machines offers tangible benefits to face identification accuracy in important applications. These results offer an evidence-based roadmap for achieving the most accurate face identification possible.

Job done!