Denver, CO, U.S.A. – Rank One Computing (ROC) announced the latest improvements in their industry leading Software Development Kit (SDK) for face recognition and computer vision analytics!
The ROC SDK version 2.3 brings a series of algorithm enhancements that benefit integration partners in a wide range of use-cases, including ID proofing, video analytics, and forensics.
No improvement is more important than the new Presentation Attack Detection (PAD), or “facial liveness”, solution. With v2.3, ROC reduced the error rate of its patented, single-frame passive liveness method by 2x to 5x across different spoof mediums and genuine user presentation categories. In a range of scenarios, ROC is now measuring better than 99% PAD accuracy. These exciting improvements will better serve our partners who are developing identity proofing and access control systems. ROC expects to deliver even more improvements to liveness in early 2023.
In this release, ROC is also delivering a tremendous accuracy improvement to its tattoo recognition algorithm. The new algorithm deploys cutting edge advancements in convolutional neural networks and machine learning to reset industry expectations for tattoo recognition capability. When comparing ROC v2.3 Tattoo to the last published U.S. NIST Tattoo Recognition Technology Evaluation (Tatt-E) benchmark in 2018, the current ROC tattoo algorithm reduces error rates by an order of magnitude on similar law enforcement imagery when compared to the previous best reported algorithm. This radically improved solution effectively makes ROC the world’s most accurate tattoo recognition algorithm provider. As a result ROC SDK v2.3 will provide game-changing improvements to forensic investigators working to identify deceased individuals and solve major cases.
Another exciting feature is the overhauled version of ROC Facial Analytics. With v2.3 a new standalone algorithm can accurately estimate dozens of facial analytic features in less than 50 milliseconds on a single CPU core. Even at this speed, v2.3 still delivers mask detection, ISO/IEC 19794-5 and ICAO 9303 portrait quality compliance checks, facial pose estimation, age estimation, and a wide range of other analytics. Also new in this release, is the ability to help construct an ICAO compliant image by automatically cropping the face and removing the background.
Finally, as ROC continues to improve our license plate recognition capabilities, v2.3 delivers another round of enhancements to this solution with improved plate detection across a broad range of conditions.
For a more detailed set of information about all of these algorithmic capabilities please see our recent technical article.
With each new release, the ROC SDK continues to raise the bar for software libraries and APIs that provide facial recognition and video analytics, all while maintaining lightning-fast speed and industry leading accuracy and efficiency. This is why ROC.ai is trusted by an ever-growing number of users and partners in Fintech, Security, Government, and other industries.
Contact us today to learn more about ROC SDK v2.3 and other ways Rank One can support your organization!
About ROC: Rank One Computing (ROC) is the most trusted provider of Facial Recognition (FR) algorithms to the U.S. Military, Law Enforcement, Fintech, and Commercial organizations. We are employee-owned, ethics driven, and 100% Made in America. Our AI/ML computer vision algorithms lead the industry in security, accuracy, and speed as proven in NIST government testing, tactical military applications and hundreds of millions of identity proofing transactions. ROC offers a growing suite of mobile and desktop computer vision solutions for all your identification and analytics needs. Our company has offices in Denver, Colorado and Morgantown, West Virginia.