Next-Gen Facial-Recognition May See Through Disguises
Even as facial-recognition technology continues to improve by leaps and bounds, up until now it has been unable to work around simple disguises such as fake beards, head coverings, and glasses. But that could all be changing. Engineers at India’s National Institute of Technology and Science and the U.K.’s University of Cambridge have developed a next-generation facial recognition framework that can identify people even when they’ve actively sought to obscure their faces. According to researcher Amarjot Singh, “Disguise Face Identification is an extremely challenging and interesting problem that is of great interest to law enforcement.” The deep learning-based system works by identifying 14 key areas of the face, including 10 for the eyes, 3 for the lips, and one for the nose, which the system is capable of estimating even when they are obscured. In early tests, the system was accurate 56 percent of the time when dealing with head coverings, but dropped to 43 percent when glasses were added. And while such figures may not be considered compelling in a court of law, they could signal a future where AI improves exponentially in searches and identification to assist Law Enforcement.
This video was produced by YT Wochit Tech using http://wochit.com