Kneron facial recognition software unlocks entry-level smartphones in 0.2 seconds
Artificial intelligence firm Kneron reports that its facial recognition solution unlocks entry level and mid-range smartphones in 0.2 seconds, with an error rate lower than one in ten thousand.
The technology works on any device that has a front-facing phone with more than a million pixels, applying a deep learning algorithm which can recognize multiple faces from a single device, and also protects against photo and video spoofs.
According to the announcement, Kneron Real Facial Recognition solutions include three technical versions: optical sensing analysis, biological behavior recognition, and background depth detection. Optical sensing analysis can determine whether a face is three-dimensional through analyzing the lighting variations on the face; biological behavior recognition integrates behavioral detection, such as eye blinking, to determine whether a face is real; the background depth detection solution can be used if there is an autofocus function in a smartphone’s front camera. It can analyze the scene depth of field to avoid unlocking by a two-dimensional image.
“Traditional facial recognition technology still has a number of security holes,” Kneron Chairman of the Board Chun Chen Liu explained. “For example, in order to carry out recognition to determine whether photographs or videos are real people, high-end hardware specifications are needed, such as infrared lenses, high pixel lenses, and even altered hardware designs. Kneron provides innovative facial recognition solutions which endow entry-level cellphones with the capacity to utilize facial recognition unlocking functions which do not require new cell phone designs or new hardware components. Even through just software upgrades, they make it possible to achieve high accuracy and highly secure facial recognition.”
Facial recognition solutions from Kneron are already being integrated by mobile phone manufacturers, and the company is expanding into the global market through a partnership with GMobi.
Kneron also recently announced the development of a series of neural processing unit (NPU) chips to enable face and gesture recognition on drones, smartphones, and IoT devices.