It sounds like a dystopian nightmare. But in the near future, cities throughout the world could be equipped with networks of cameras analyzing facial features of pedestrians for certain traits that researchers believe are predictive of criminal behavior.
A pair of Chinese researchers say they’ve come up with a computer program that uses algorithms to effectively analyze facial features and detect whether an individual is a criminal.
In testing their program, the researchers ran the photo ID images of hundreds of Chinese males between the ages 18 and 55, a portion of whom had been convicted of violent crimes “including murder, rape, assault, kidnap, and robbery, through the program.
Using a set of classifiers, the algorithm was reportedly able to pick out the convicts with nearly 90 percent accuracy.
“All four classifiers perform consistently well and produce evidence for the validity of automated face-induced inference on criminality, despite the historical controversy surrounding the topic,” the researchers write. “Also, we find some discriminating structural features for predicting criminality, such as lip curvature, eye inner corner distance, and the so-called nose-mouth angle.”
“By extensive experiments and vigorous cross validations,” the researchers add, “we have demonstrated that via supervised machine learning, data-driven face classifiers are able to make reliable inference on criminality.”
While it would take far more testing before software like this became usable by government law enforcement agencies, that it’s being considered by scientific researchers underscores the importance of promoting civil liberty and limits on how and where government is legally able to use facial recognition.
The post Researchers claim computer can point out criminals based on facial features appeared first on Personal Liberty®.