Ever AI Leads All US Companies on NIST’s Prestigious Facial Recognition Vendor Test

A leader in the West finally emerges to challenge China and Russia in face recognition accuracy

SAN FRANCISCO, Nov. 27, 2018 (GLOBE NEWSWIRE) -- Ever AI, a face recognition platform designed specifically for mission-critical applications, today announced that its facial recognition algorithm has ranked as the most accurate among all United States companies, and tied for first globally in the Wild Images category, in the latest Face Recognition Vendor Test (FRVT) 1:1 evaluation conducted by the U.S. National Institute of Standards and Technology (NIST). The company also ranked as the top private US-based company across 28 categories in the first ever NIST FRVT 1:N evaluation.

In addition to the company’s current world-best score of 99.85% in the University of Massachusetts’ Labeled Faces in the Wild (LFW) benchmark and its top score of 99.04% in the University of Washington’s MegaFace benchmark, the NIST FRVT results further establish Ever AI’s leadership in delivering highly accurate, highly secure, mission-critical face recognition - handily beating out more established providers like NEC, Idemia, Hikvision, Cognitec, Gemalto, Panasonic, Toshiba, Vocord, AnyVision, Rank One Computing, Realnetworks, Camvi Technologies, Sensetime and Megvii (Face++).

“These independent evaluations establish us as the de facto leader of mission-critical face recognition in the US,” said Doug Aley, CEO of Ever AI. “Our flexible deployment options, accuracy at a variety of distances and poses, diverse training data sets that ensure low bias, and our custom customer models that dramatically reduce the risk of adversarial attacks, make our platform the best option for authentication, access control, security and surveillance applications.”

The NIST FRVTs provide wholly independent evaluations of commercially available and prototype face recognition technologies. The FRVT 1:1 evaluation focuses on one-to-one use cases like face verification for authentication and access control, while the FRVT 1:N evaluation assesses one-to-many uses cases like face identification across enrolled galleries containing at least 10 million identities. The results are used by the U.S. Government and law enforcement agencies in determining where and how facial recognition technology can best be deployed, and by the global security and financial services industries as the gold standard in determining face recognition accuracy.

The NIST FRVT results come on the heels of the release of Ever AI’s new liveness detection offering, which helps financial services and other security-centric companies protect against presentation-based attacks and other sophisticated spoofing techniques. Liveness detection was added to the complete suite of Ever AI APIs and SDKs that offer detection, verification, identification, clustering and attribute extraction (age, gender, race, emotion, landmarks and occlusions). The company’s SDKs support a variety of technology configurations and come with models as small as 4MB, perfect for processing at the edge. APIs are also available for on premises, self-hosted deployments.

To learn more about Ever AI’s mission-critical face recognition platform and SDKs, visit: http://ever.ai.

About Ever AI
Ever AI delivers a higher standard of face recognition. Trained on an ever-expanding private global dataset of 13 billion photos and videos, Ever AI’s technology is designed specifically for Enterprises who need mission-critical face recognition that excels at speed & accuracy, provides superior levels of security, and can be deployed in any environment. Notable customers include: SoftBank Robotics and Bluescape.

Founded in 2013, Ever AI is headquartered in San Francisco, California. The privately held company is backed by some of the best enterprise and AI investors in the world including Khosla Ventures, Icon Ventures, Felicis Ventures, Transmedia Capital and SV Angel. For more information, please visit: https://ever.ai/.

Media Contact
Michael Walton
E: press@ever.ai