Performance

The facial recognition algorithms are based on pattern matching in collaborative situations and on machine learning in non-collaborative situations. The combination of the two algorithms allows to reach very high levels of FAR (False Acceptance Rate) and FRR (False Rejection Rate).

Face DB Dimension: Unlimited

Face Searching speed in pre enrolled DB: 1:N <200ms

Face Enrolling speed: <100ms

Max. Numer of camera in the system: Unlimited

The Facial recognition algorithm is verified by NIST (USA National Institute of Standard and Technology 2019) with FNMR (False Non Match Rate) 0.021.
It works by allowing an offset up to over 25 degrees of the face of the subject to be recognized and handles more than one face for the same subject. It is also able to find subjects framed in profile when the sample is available.

LFW** Faces 99,8%
University of MA 2017

FNMR* 0,021
NIST (National Instistute of Standard and Technology 2019)

The suite is able to compare a face taken from a frame extracted from an IP video surveillance camera with faces included in a database containing millions of pre-processed photos in less than a second.
It has a client-server architecture in which there is no technical limitation on the number of cameras, number of client workstations and number of manageable faces from the system.

The suite has a client-server architecture in which there is no technical limitation on the number of cameras, number of client workstations and number of manageable faces from the system. The suite is able to manage video streams both from 2D cameras, both from 3D cameras. The combination of the two flows makes it possible to use the antispoofing algorithm that exceeds the problem of using a photograph instead of the real face.

The suite combines and correlates the following entities: Person/Vehicle/Object to identify, points of recognition events, actions in a circular manner. It supports an unlimited number of recognition points and an unlimited number of entities: person, vehicle, object, an unlimited number of client and user applications and an unlimited number of faces in the database.

The suite combines and correlates the following entities: person/vehicle/object to identify, points of recognition events, actions in a circular manner. It supports an unlimited number of entities: person, vehicle, object. Supports an unlimited number of client and user applications and an unlimited number of faces in the database.