AI Matches Millions of Faces in Real Time
SEATTLE, July 17, 2018 /PRNewswire/ -- RealNetworks,
Inc. (NASDAQ: RNWK), a leader in digital media software and
services, today announced SAFR™, a highly accurate, machine
learning facial recognition platform, architected to economically
scale with high performance and rapid processing to detect and
match millions of faces in real time. Utilizing machine learning,
SAFR – Secure Accurate Facial Recognition – is continuously
improving efficiency, accuracy, and reach.
SAFR is distinguished from other facial recognition platforms in
three ways:
- World class accuracy — SAFR recognizes faces with proven 99.8
percent accuracy for Labeled Faces in the Wild (LFW), based on the
University of Massachusetts benchmark.
In addition, the June results of RealNetworks' first submission to
the National Institute of Standards and Technology (NIST), ranked
SAFR's recognition algorithm for "Wild Faces" False Non-Match-Rate
(FNMR), as the best of any platform submitted by a United States-based company and as one of the
world's top recognition algorithms. NIST is the premier measurement
standards laboratory for the United States.
- Extraordinary efficiency and flexibility — SAFR works
seamlessly with existing IP-based cameras and readily available
hardware to recognize people in real time, helping to enhance
secure access and surface insights. SAFR supports both cloud and
local storage. System integrators and application developers can
easily integrate with the SAFR platform through RESTful API's, an
SDK, and dashboard.
"The architectural design of SAFR
takes a novel approach to facial recognition at scale," said
David Sobeski, former chief
technology officer, and senior vice president, Walt Disney Company.
"This makes SAFR commercially viable for a wider range of practical
real-world applications."
- Focus on privacy and socially positive use cases — SAFR
encrypts all facial data and images to ensure privacy. When used
locally, no personal or facial data is ever transmitted over the
Internet. Moreover, RealNetworks is announcing that SAFR is being
made available for free to over 100,000
K-12 schools in the United
States and Canada (see
separate release).
"Privacy concerns are paramount in
our design thinking," said Reza
Rassool, CTO of RealNetworks. "SAFR can be deployed securely
in the cloud or purely on-premises, where all face and event data
remains on site. In both cases private data is encrypted in transit
and at rest."
SAFR supports numerous secure access use cases where facial
recognition can replace the use of an ID badge, securely automate
entry to facilities, trigger notifications, and log events for
analytics. SAFR provides facial detection and tracking of many
faces in a single camera feed. Each face can be selectively
analyzed for age, gender, sentiment, and liveness. Faces are
rapidly matched to a database of enrolled faces, returning each
identity with a recognition confidence score. SAFR works with
off-the-shelf IP cameras and runs on mobile devices and readily
available computers.
Lasse Andresen, ForgeRock's chief
technology officer, said, "At ForgeRock, our focus is to enable
organizations of all sizes to build trusted digital relationships,
and nothing can be more notable than protecting the bond between
schools and its students. We congratulate RealNetworks in
delivering the SAFR platform and look forward to working together
to advance this exciting initiative."
Building on its legacy and leadership in streaming
media—including compression, media delivery, digital rights
management, and efficient scalability—RealNetworks' expertise now
extends to AI, machine learning and deep neural networks: the
foundation of SAFR.
SAFR for K-12 is being offered for free to K-12 schools in
the United States and Canada. SAFR will be made available through a
commercially-paid license for other industries. More information on
SAFR is available at www.realnetworks.com/safr/.
About RealNetworks
RealNetworks creates innovative
applications and services that make it easy to connect with and
enjoy digital media. RealNetworks invented the streaming media
category and continues to connect consumers with their digital
media both directly and through partners, aiming to support every
network, device, media type and social network. Find RealNetworks'
corporate information at www.realnetworks.com.
RealNetworks® and the company's respective logos are trademarks,
registered trademarks, or service marks of RealNetworks, Inc. Other
products and company names mentioned are the trademarks of their
respective owners.
To learn more, contact public_relations@realnetworks.com
View original content with
multimedia:http://www.prnewswire.com/news-releases/realnetworks-launches-safr-a-best-in-class-facial-recognition-platform-300681916.html
SOURCE RealNetworks, Inc.