(by Chris Burt | Biometric Update) – A pair of new features have been added to Clearview AI’s facial recognition capabilities to help law enforcement identify people of interest in investigations more easily if their faces are covered or obscured.
Clearview CEO Hoan Ton-That told Wired that the company has added machine learning tools to “deblur” pictures, or sharpen blurriness, and depict the full faces of people wearing masks.
MIT Professor Aleksander Madry, who specializes in machine learning, told Wired that he suspects the accuracy of Clearview’s new facial recognition tools will be poor, and they will introduce unintended bias.
“Any enhanced images should be noted as such, and extra care taken when evaluating results that may result from an enhanced image,” Ton-That cautions.
The company has also reached 10 billion images scraped from the web, Ton-That says, three times as many as previously reported. He also claims that Clearview is up to 3,100 law enforcement and government customers for its face biometrics-matching service.
Ton-That also said that while he thinks more people support the use of facial recognition in law enforcement than oppose it, it is good that those opposing the biometric application are vocal, so their concerns can be addressed.
The company recently lost a bid to halt legal proceedings against it on constitutional grounds for allegedly violating biometric data privacy laws.