Facebook’s DeepFace as Accurate as Humans

DeepFace- Facebook’s New Face Recognition System May be as Accurate as Humans

A new complex artificial intelligence system has now been developed by Facebook, which may detect faces as accurately as humans. Called DeepFace, this new facial recognition system seems to have broken all conventions of earlier versions and is capable of rendering a 3D model of any human face based on its availability on Facebook.

DeepFace is believed to correctly identify a face from two different photos with a success rate of a whopping 97.25%, which is commendable considering other face recognition systems.

Facebook supposedly uses an algorithm called ‘Deep Learning’ to render a 3D model of a human face, and then uses over 120 million parameters to analyse the image, before detecting it.

Facebook Launches New Face Recognition System that Detects as Accurately as Humans

The researchers detailed the abstract as follows- ”In modern face recognition, the conventional pipeline consists of four stages: detect => align => represent => classify. We revisit both the alignment step and the representation step by employing explicit 3D face modeling in order to apply a piecewise affine transformation, and derive a face representation from a nine-layer deep neural network.”

While this new technology is still in its developmental phase, it seems Facebook is now attempting to strengthen its security and privacy features. DeepFace, with its stunning new technology, seems to have blurred the barrier between humans and technology, and

“The social and cultural implications of face recognition technologies are far reaching, yet the current performance gap in this domain between machines and the human visual system serves as a buffer from having to deal with these implications,” the researchers explained. DeepFace ”has closed the majority of the remaining gap.”

DeepFace- Facebook’s New Face Recognition System May be as Accurate as Humans