RealNetworks’ SAFR facial recognition tech now available in China
RealNetworks has signed a strategic cooperation agreement with Shenzhen Homtar High-Tech Co. to bring its AI-based SAFR platform and facial recognition technology to China. SAFR is designed to identify faces in real-world conditions, including people in motion, in dim lighting, and at occluded angles.
According to a company statement, SAFR can reliably match against millions of faces in under a second and can handle a variety of conditions, including face makeup, hats, glasses, varying hair style, and aging. It also has a high tolerance to facial yaw, tilt and rotation, minimizing errors by allowing error rate adjustment.
According to NIST, RealNetworks’ computer vision achieved a high performance rating with an accuracy score of 0.048 for Wild Faces FNMR. In a test developed by the University of Massachusetts, SAFR achieved a 99.8% accuracy score for “labeled faces in the wild” (LFW).
SAFR works with existing IP cameras and available hardware to match faces in real time, enhancing secure access. SAFR supports cloud and on-premises local storage. System integrators and application developers can integrate with the SAFR platform through RESTful APIs, SDKs and dashboards. SAFR encrypts all facial data and images to ensure privacy and when used locally, no personal or facial data is transmitted over the internet.
“RealNetworks uses the latest computer vision, machine learning theory and technical methods to develop the industry-leading SAFR software,” explains RealNetworks’ CTO Reza Rassool. “We achieved excellent results by training for faces in photos and videos from real-life video. We are very happy that we can work together with Homtar to commercialize this technology in China.”
Homtar will launch the technology with its partner Shenzhen Wiseme Electronics Co. “We’re excited that RealNetworks’ advanced SAFR technology will be promoted and operated in China,” commented Ms. Lin Luhua, Chairman of Homtar Corp. “Through the development of localized applications, we will bring safer and more efficient products to our partners. The first trial system will be launched in Shenzhen with Kingbrother Tech Ltd. within one month.”
According to the partnership announcement, RealNetworks and Homtar will work together to develop a series of SAFR commercial applications including VIP identification, retail analytics, secure area monitoring, live data collection, visitor IDs and screenings, and digital signatures.
Since rolling out SAFR this summer for free to grades K-12 schools in the United States, RealNetworks has launched its facial recognition product in India and has signed strategic partnerships with firms in Latin America and Japan.
In October RealNetworks released its Best Practices guide for using facial recognition technology to support safer K-12 campuses.