Face detection and recognition algorithm, rigorously trained over a decade on the most challenging image data
AppliedRecognition’s family of products, from authentication to photo tagging, are all built on the FaceLocate algorithm.
Built over a decade, FaceLocate has been trained on face detection, liveness detection, and face recognition on hundreds of thousands of, primarily difficult and lower quality images from photo tagging.
Tackling challenging datasets has improved the accuracy of FaceLocate specifically for authentication. Operating under webcam conditions for device log in creates specific challenges for face recognition vs comparing two mug shots, for example illumination and slightly different poses. It’s critical for face authentication applications that these variables are incorporated into the overall training of the system.
Questions on face authentication accuracy?
The industry has adopted a testing standard against a dataset known as the FERET database. FaceLocate’s equilibrium balance point between false acceptance and false rejection is 0.2 % for the FERET database. FaceLocate’s flexible parameter settings allows for higher security or greater ease of use, with a 99.99 % security level easily achievable.