TL; DR: The SAFR platform from RealNetworks is often a premier facial recognition option for live video. Power by AI, the solution is optimized pertaining to accuracy and performance by having a marathon-like approach to product development aimed at continuous improvement. Designed to accommodate numerous environments, from enterprises for you to K-12 schools, SAFR is providing new safety opportunities for you to customers worldwide.
Should you have listened to music as well as watched a video online in the last decade, chances are you’ve applied RealPlayer, one of the 1st applications for streaming media over the web.
The software, formerly generally known as RealAudio Player, was unveiled in 1995 by RealNetworks. Right now, as a multimillion-dollar firm publicly traded on NASDAQ, RealNetworks remains at your forefront of innovation online.
“Our 25-year legacy as being a pioneer in the online video space has led us to make the world’s foremost podium for facial recognition throughout live video, ” explained Dan Grimm, Vice Us president and General Manager involving Computer Vision at RealNetworks. “Secure, Accurate Facial Recognition (SAFR) is surely an AI-powered product that supplies next-generation visibility and situational recognition for security professionals. ”
Unlike various other facial recognition platforms, SAFR was meant to accurately identify faces which have been in motion, under inadequate lighting conditions, partially obscured, or misaligned — real-life traits that cameras might capture in a very natural setting.
SAFR also employs an integrated set of technologies to mix high performance with a new flexible architecture. This approach allows your forward-thinking company to support various practical use cases in enterprises and K-12 educational facilities alike, from secure gain access to and gateway safety for you to VIP loyalty and locale monitoring.
Ultimately, RealNetworks is dedicated to building the world’s nearly all trusted facial recognition podium through SAFR, leveraging ongoing improvements throughout both artificial intelligence and machine finding out how to continually raise the bar regarding accuracy and performance.
Using AI to generate Sense of the Entire world Around Us
As much even as we fear a technological takeover involving sci-fi movie proportions, we will need to face the facts: Cameras have outnumbered humans for this earth. The current world inhabitants is estimated at over 7. 7 billion, and there are about 14 billion cameras today. By 2022, that variety is predicted to go up to 45 billion. With nearly six cameras for all on earth, many humans will never view almost all of the visuals captured by the product — there is simply too much data for the neural to process.
“Instead, artificial intelligence will visit the edge and make sense of so much visual information, ” Dan said. “With SAFR, we will enable customers in several verticals in the stability space — home, non commercial, enterprise, airports, hospitals, stadiums — to leverage our platform to be aware of the world around these people. ”
RealNetworks comes with a software development kit (SDK) for SAFR that could enable manufacturers of IoT devices inside edge computing space for you to leverage its world-class skin recognition technology through are living video. This will help, for example, the head of security at the hospital or stadium for you to derive new capabilities and insight from numerous live video feeds in a way that was humanly impossible during the past.
“Device manufacturers can license our technology to create truly world-class computer eye-sight and facial recognition on the edge, ” Dan explained. “We have SDKs that could run on devices including video conferencing cameras, home security cameras, mobile devices, chillers, drones, cars, etc. ”
An increased Level of Accuracy along with Low Propensity for Error
Dan told us that will RealNetworks has observed remarkable advances in deep mastering with convolutional neural networks (a sort of artificial neural network employed to analyze images) over the last few years, and worked as a chef to stay one step ahead inside quickly evolving space.
When assessed resistant to the benchmark of Labeled Faces inside Wild (LFW), a database of photographs put together by the University of Massachusetts to analyze unconstrained facial recognition underneath real-world conditions, SAFR exhibited 99. 87% accuracy. Moreover, RealNetworks continually submits its algorithms on the National Institute of Criteria and Technology (NIST) pertaining to performance evaluations.
NIST’s test results via April 2019 ranked your SAFR algorithm as the two fastest and most stream-lined among algorithms for untamed images, with a false non-match rate (the rate from which an algorithm mistakenly identifies two images in the same individual as different) of below 0. 025. In terms of template extraction — the task of creating a skin signature — NIST ranked SAFR’s algorithm one of several top seven worldwide.
“We are dedicated to creating a computer vision platform that may be truly excellent, exhibiting high levels involving accuracy and low levels of bias regarding variances in skin firmness and gender — these are generally things that we are generally continuously improving upon, ” Dan said. “We believe this technological innovation offers tremendous benefits but has got to meet a certain bar of excellence for individuals to retain customer have confidence in. ”
Increase School Safety totally free in K-12 Schools
RealNetworks introduced SAFR on the market in 2018 as being a free download for K-12 schools in the us and Canada looking to further improve security and convenience for trusted members in the community. After guest complete a registration process with a tablet with staff endorsement, the technology will match faces instantly to streamline the accessibility and check-in process.
“As a parent or possibly a staff member, you can register yourself inside lobby of the institution, ” Dan said. “The next occasion you come to top gate of the campus, as an alternative to standing there and smacking a buzzer that someone may or might not exactly answer, with or without expertise in who is there, a camera will acknowledge you and automatically unlock the threshold. ”
SAFR at a new schoolThe AI-powered solution determines faces in real-world circumstances.
In the same way that this technology makes security-conscious campuses offered to trusted members in the community, it may also be configured to realize persons of concern — including expelled students or ex – employees — from within and outside of the school. In this means, the platform adds real value for American schools, which have been overwhelmed with deadly violence within the last few few decades.
“We support our customers, particularly inside security space, see what matters in the real world, ” Dan said. “We get begun with facial identification, but we’ve also just lately launched with person detection all of which will continue to expand our vision offering to cope with the needs of each of our customers for visual intelligence inside spaces they care with regards to. ”
A Marathon Procedure for Improving Accuracy
Dan said building remarkably accurate facial recognition software is often a marathon, not a race. In addition to following feedback from the education and learning and enterprise security areas, RealNetworks strives to continuously help the accuracy and speed of its algorithms to be at the forefront in the industry.
Ultimately, the mission at SAFR and RealNetworks should be to build the world’s nearly all trusted visual recognition podium. “In order to meet that responsibility, we believe we will need to design and develop a totally excellent product, ” Dan said.