New evaluation against NIST dataset results in highest ever accuracy for AimFace
London, 13 January, 2019: AimBrain, the Biometric Identity as-a-Service platform, announced the results of its latest benchmark on NIST Color FERET version 2 Dataset.
The dataset is widely regarded as the main benchmark in the biometric industry, for a consistent and unambiguous comparison of authentication vendors. Its dataset is used to develop, test and evaluate facial recognition algorithms.
AimBrain’s facial authentication module AimFace achieved an all-time low equal error rate (ERR) of 0.03%, equating to a recognition accuracy of 99.97%. The equal error rate is the point at which false accepts equals the number of false rejects, and the lower the EER, the better the technology at recognising legitimate users and stopping fraudulent ones.
The results coincided with Find Biometrics’ Year In Review Survey 2018, which confirmed that the most exciting developments in biometric authentication technology were deemed to be those in facial authentication, voted for by 37% of respondents.
For more information, download the AimFace NIST whitepaper.