Rank One, a facial recognition SDK provider, recently announced a periocular recognition algorithm. A periocular approach to facial recognition uses images of the eye and the eyebrow area to perform facial recognition. This new algorithm is especially timely given the COVID-19 outbreak which limits access to biometric approaches that require touch (e.g. fingerprints) and full facial recognition now that much of the population is currently wearing masks.
In testing, Rank One has already achieved accuracy rates similar to complete facial recognition results from 2018. Further, it already outperforms many of the NIST FRVT face recognition algorithms currently used throughout national security and critical infrastructure facilities. While the periocular algorithm is not a fast as Rank One's flagship facial recognition algorithm, the company plans on getting it up to speed shortly. The company rushed the release in order to get a product to market that can actively meet needs during the COVID-19 pandemic.
Public documentation for the new algorithm is not currently available. Those interested should reach out to the Rank One team. Using the new algorithm is similar to using the facial recognition algorithm in the existing ROC SDK. The periocular algorithm is available with the standard ROC SDK license.