Researchers find combination of human experts and algorithms most accurate for facial recognition
A team of scientists from NIST and three universities has found that trained experts in facial recognition make more accurate judgements with a computer partner than a human one, or than top-performing algorithms on their own, in a study combining forensic science, psychology, and computer vision research.
The researchers undertook to the most comprehensive examination of face identification performance across a large and varied group of people to date, according to a NIST announcement, in response to a National Research Council report from 2009, “Strengthening Forensic Science in the United States: A Path Forward.” Their findings are published in the journal Proceedings of the National Academy of Sciences.
“This is the first study to measure face identification accuracy for professional forensic facial examiners, working under circumstances that apply in real-world casework,” said NIST electronic engineer P. Jonathon Phillips. “Our deeper goal was to find better ways to increase the accuracy of forensic facial comparisons.”
Phillips also said that three years ago the best algorithms performed comparably to untrained students, but are now just as effective as highly trained professionals.
Human examiners working together achieved better performance than individuals, but the best results were delivered by human-computer cooperation.
“If combining decisions from two sources increases accuracy, then this method demonstrates the existence of different strategies,” Phillips said. “But it does not explain how the strategies are different.”
Researchers from the University of Texas at Dallas, the University of Maryland, and Australia’s University of New South Wales contributed to the study.