“According to the KAIST researchers, the system allows for facial recognition with 97 percent accuracy, using only 0.62mW of power.”
A research team at the Korea Advanced Institute of Science and Technology’s Department of Electrical Engineering has developed a new biometric AI system in collaboration with domestic startup UX Factory Co.
The system has two key components. One is a special semiconductor chip called CNNP (‘CNN Processor’) designed specifically to run AI algorithms with minimal power consumption; the other is facial recognition software, called K-Eye, developed specifically for the CNNP chip. According to the KAIST researchers, the system allows for facial recognition with 97 percent accuracy, using only 0.62mW of power.
As for its physical form, the researchers made the system into a both a wearable device and a dongle, either of which is designed to pair with a smartphone via Bluetooth, enabling the user to instantaneously capture the facial biometrics of a subject using the smartphone camera.
The KAIST team and UX Factory Co. say they’re aiming to commercialize the system by the end of this year, by which time sophisticated facial recognition could be surging in popularity if Apple incorporates the technology into its next iPhone as it is currently anticipated to do.