Real-time Identification of a Person by a Facial Image
PAPILLON POLYFACE-3 is adapted as much as possible to address challenging issues of automatic real-time identification of a person from a facial image acquired with a mobile or stationary surveillance camera (express identity checking). It is to this end that POLYFACE provides the following functionality:
Matching Algorithms
POLYFACE-3 uses proprietary search algorithm 3DiVi Face SDK (developed by 3DiVi company belonging to PAPILLON’s group). It is based on the method of trainable neural networks and provides the high indices of accuracy and reliability in automatic comparisons.
Since the start of 2017, our algorithm has been participating in the NIST* Face Recognition Vendor Test (FRVT). According to the published results, it demonstrates the highest efficiency, being now the fastest in the world and surpassing manyfold its best analogues in comparison speed.
* Results shown from NIST do not constitute an endorsement of any particular system, product, service, or company by NIST
(https://www.nist.gov/programs-projects/face-recognition-vendor-test-frvt-ongoing).
System Architecture
The architecture of POLYFACE-3 is based on the client-server model that allows POLYFACE workstations to independently request the server’s functions and services. The search process is based on the technology of distributed computing.
Access to the PAPILLON POLYFACE-3 database is available through:
Integration with PAPILLON AFIS
Integration with PAPILLON AFIS involves automatic transfer of data associated with every newly entered tenprint (demographics, mugshots and other photo images, verbal description of an individual's appearance and traits) to POLYFACE-3 database for enrollment and matching. Results of a search are transferred back to the AFIS site.
Alternatively, the POLYFACE-3 program modules can be integrated into PAPILLON AFIS software architecture to enable multibiometric enrollment and identification.
|