New Rank One Embedded algorithm triples speed with smaller footprint

Rank One Computing has released an accelerated version of its standard ROC SDK algorithm called ROC Embedded for mobile devices and video processing applications that is says increases speed by 300 percent, with enrollment speeds on a single CPU core with a template size of 85 bytes of less than 30 milliseconds.

The new Rank One algorithm also has a memory footprint of only 7MB and is a faster algorithm for use in embedded devices and video applications. The algorithm corresponding with the company’s ROC SDK version 1.17 had higher accuracy in recent NIST FRVT testing than the top algorithm from April, while performing at a high speed.

Rank One has focused on delivering leading speed and efficiency while maintaining top-tier accuracy, and the latest release of ROC Embedded redefines the possibilities of mobile, embedded device, and video processing applications, according to the announcement.

Article Topics

algorithms  |  biometrics  |  facial recognition  |  mobile device  |  ROC