Forensic examiners with experience of comparing face images for law enforcement and government agencies still outperform untrained participants and computer algorithms, according to a major study released today.A research team led by Dr David White from the University of New South Wales has found that fusing responses of multiple experts produced “near-perfect” performance.”Computers at the moment can only outperform people in tasks when the images are of high quality, but often they are not”, says White in the study. “I don't think there will ever be a time when humans are not needed in this process.”The study, conducted by White and his colleagues from the National Institute of Standards and Technology and the University of Texas at Dallas, was published this week in the Proceedings of the Royal Society B journal.The researchers said that the examiners' superiority was greatest at longer durations.”The examiners' superiority was greatest when they had a longer time to study the images, and they were also more accurate than others at matching faces when the faces were shown upside down. This is consistent with them tuning into the finer details in an image, rather than relying on the whole face.”To produce the study, the group tested 27 experts who were trained according to standards established by the Forensic Identity Scientific Working Group, with each expert given either two or 30 seconds to decide if the images were of the same person or not.”On the tests most reflective of their daily work they made 7 per cent errors, compared to untrained student performance where it was about 15 per cent,” White's team wrote. “To the best of our knowledge, this is the first convincing demonstration of a professional group showing higher accuracy on face matching tasks.”In one of the smaller tests, examiners and controls had to match some pairs of photos that facial recognition software had got 100 per cent wrong.

Also in the news:

Default ThumbnailTECH5 opens subsidiary in the UK Default ThumbnailSmilepass launches Nigeria project Default ThumbnailApple patents heart rate sensor in watch to identify owner