ROC posts best latent fingerprint matching speed in NIST test with new algorithm

A new latent fingerprint matching software development kit has been launched by ROC, with a high score in the NIST Evaluation of Latent Fingerprint Technologies (ELFT) and what the company says is 500-times faster performance than the industry standard.

The U.S.-based company’s first-ever ELFT entry was found to be in the top three among global biometrics providers for accuracy, based on an aggregation of scores across various datasets and accuracy metrics for both distal prints and palm prints, according to an announcement. The average false non-identification rate (FNIR) across all distal evaluation datasets of the “ROC+005” algorithm, which is part of the ROC SDK 3, was 0.1976. The company’s algorithm also completed matches in 15 seconds, more than 150 times faster than the next-fastest provider.

A fingerprint segmentation algorithm from ROC also performed well in a NIST evaluation last year.

ROC (also known as Rank One Computing) provides its facial recognition, liveness detection, iris and fingerprint biometrics all within the SDK. The third-generation API also includes face analytics, license plate recognition and object detection.

In addition to their performance and speed advantages, ROC says its algorithms are lightweight enough for deployment at the network edge, opening up new potential applications in law enforcement, border security and counterterrorism. The company specifies possible use cases such as real-time checks against police databases and missing persons lists during traffic stops.

“Algorithms have gotten very good at matching high-quality fingerprints, but matching latent or partial fingerprints has been a persistent challenge that typically requires 45+ minutes,” says Scott Swann, CEO of ROC. “The fact that our algorithm makes incredibly accurate latent matches in just seconds opens a whole new world of applications. Just think about the importance of speed and accuracy at border crossings as we begin to realize some of these new opportunities.”

ROC Senior Scientist Dr. Joshua Engelsma says the advances in SDK 3 are due to a combination of the latest developments in computer vision and machine learning with established techniques. He expects “substantial improvements to our already top-tier latent fingerprint matcher in the very near future.”

Dr. Brendan Klare, ROC Co-founder and Chief Scientist, says, “we’re excited to push the state of machine-learned CV algorithms to other modalities beyond face recognition. Significant advancements in fingerprint technology have been few and far between. We are now turning that tide thanks to the efforts of our new fingerprint center of excellence in Grand Rapids, Michigan. We look forward to working closely again with fingerprint experts across the globe to bring these impactful capabilities to our customers.”

ROC’s software was integrated into Yoti’s identity verification platform at the close of 2023.

Article Topics

biometric matching  |  biometric testing  |  biometrics  |  ELFT  |  fingerprint biometrics  |  NIST  |  ROC  |  SDK