VOCORD increases recognition accuracy of VOCORD FaceControl biometric facial identification system


Moscow. September 29, 2016. The Russian Vocord Company, one of the leading developers and producers of intelligent video surveillance and audio logging systems (a portfolio fund company with stakes owned by RVC’s S-Group Ventures and Leader Innovations; a Skolkovo Foundation resident) has increased the recognition accuracy of the VOCORD FaceControl biometric facial identification system.

The latest VOCORD neural-network algorithm, recognized as the best worldwide in accordance with MegaFace test results, shows a four percent increase in recognition accuracy and a one million time reduction in system false positive errors (false acceptance rate - FAR). Therefore, to date, VOCORD FaceControl system facial recognition accuracy is in excess of 99 percent. (false recognition rate – FRR – less than one percent).

Furthermore, recognition accuracy in analysis of non-frontal facial images has been substantially increased (turn angle up to 45 degrees) with no increase in the percentage of false positives.

“With such high accuracy, the Vocord facial recognition algorithm can be applied in cases where high recognition accuracy is critical, particularly for access to personal data, accounts or confidential information. Further, the algorithm can be employed autonomously and in a “cloud” version,” says Vocord Technical Director Alexey Kadeyshvili. “Even though no system can guarantee a one hundred percent absence of errors, they can be reduced to a minimum. This is what we have shown with our algorithm.”

Интерфейс FaceControl