Tag Archives: Hal Hodson

The trouble with police, large photograph databases and face recognition technology

Hodson, Hal Police mass face recognition in the US will net innocent people. New Scientist. October 20th 2016.

https://www.newscientist.com/article/2109887-police-mass-face-recognition-in-the-us-will-net-innocent-people/

United States Government Accountability Office Face Recognition Technology: FBI Should Better Ensure Privacy and Accuracy. May 2016.

http://www.gao.gov/assets/680/677098.pdf

Had you assumed that hiring human super-recognizers to perform face recognition tasks would be less effective, less accurate and more open to bias than using technology? Think again.

In New Scientist

Face recognition row over right to identify you in the street

15:41 19 June 2015 by Hal Hodson

http://www.newscientist.com/article/dn27754-face-recognition-row-over-right-to-identify-you-in-the-street.html#.VYktFPmqpBd

Many of the clever face recognition tasks done by technology described here could be done by an experienced super-recognizer employee, and commentators concerned about privacy can go have a big cry over the things that supers can do, because you can’t legislate that people such as super-recognizers should refrain from using natural talents.

And more…..

Facebook can recognise you in photos even if you’re not looking

15:27 22 June 2015 by Aviva Rutkin

http://www.newscientist.com/article/dn27761-facebook-can-recognise-you-in-photos-even-if-youre-not-looking.html#.VYp5JPmqpBc

Facebook almost as good as humans at recognizing faces, but how would it compare with super-recognizers?

The thing that you’ve got to remember when reading headlines like this is that human super-recognizers can recognize (recently and non-recently) familiar faces at a level of performance that is very much higher than the human norm in face recognition or face memory performance. So the technology might be closing in on human face recognition ability, but how far is it from outperforming the best humans, the super-recognizers who have been estimated as one in a hundred people? My guess is that the technology isn’t even in the same postcode as supers.

Facebook almost as good as humans at recognising faces.
by Hal Hodson
New Scientist. 18 March 2014 Issue 2961.
http://www.newscientist.com/article/dn25245-facebook-almost-as-good-as-humans-at-recognising-faces.html#.Uyhag_mSx8E