Neurotechnology’s SkyBiometry improves biometric cloud face detection algorithm
Neurotechnology subsidiary SkyBiometry has launched a new version of its biometric face detection and recognition algorithm with significant accuracy enhancements and a wider range of facial attributes used it the detection process, according to a company announcement.
The latest face detection algorithm locates more faces in more varied and difficult conditions and at multiple head rotation angles, up to and including full profile. Face recognition accuracy is increased by up to five times, when set at a low False Acceptance Rate, the company says. Other algorithms within the API for facial-feature point detection, face image quality estimation, and face properties and emotion classification have also been improved.
“The new features made available in the latest version of the SkyBiometry facial algorithm were added and augmented based on customer feedback,” says Dr. Justas Kranauskas, head of the biometric research department for SkyBiometry and Neurotechnology. “More accurate face detection and face recognition algorithms continue to make our service unique, and the additional attributes will help our customers expand the scope of their SkyBiometry API applications.”
SkyBiometry can now determine race or ethnicity, and if a person is wearing a hat, or has a beard or mustache, in addition to previous capabilities for determining gender, age, and emotion, and detecting dark glasses, closed and open eyes, and other facial properties. The SkyBiometry API can be used for face detection, recognition, grouping, and attribute determination applications. These include identifying and cataloguing images for photographers and others who work with large databases, as well as in developing time and attendance software, marketing and entertainment applications.
The service is available as a free subscription for up to 5,000 monthly calls, and a database capacity of up to 1,000 trained tags (faces).
The company notes that its algorithms placed in the top ten among all vendors in NIST Facial Recognition Vendor Test (FRVT) 2018, and is in the top 15 out of 82 vendors in the FRVT Ongoing in the “wild” category.
The cloud biometrics market is expected to be worth $2.7 billion by 2024.