Neurotechnology, a provider of high-precision biometric identification and object recognition technologies, announced the release of the SentiVeillance 5.0 software development kit (SDK). This latest version of SentiVeillance incorporates the new VeriLook face recognition algorithm featured in MegaMatcher 9.0, providing five times higher accuracy in identifying full frontal faces and 10 to 15 times higher accuracy for unconstrained facial recognition than the previous release. SentiVeillance 5.0 works with images from surveillance cameras, making it suitable for a wide range of applications in surveillance, security and public safety.
SentiVeillance uses the face recognition algorithm to match face images against internal databases, such as authorized personnel or criminal watch lists. This allows a SentiVeillance-based application to trigger alerts for recognized or unrecognized faces.
"Using state of the art technology, called deep neural networks, we were able to significantly improve facial recognition accuracy, especially for unconstrained scenarios," said Ignas Namajunas, surveillance technologies research lead for Neurotechnology.
The significantly higher accuracy for unconstrained facial identification is based on a smaller False Rejection Rate (FRR) at the same False Acceptance Rate (FAR) value.
In addition to face tracking and recognition, SentiVeillance provides real-time moving object detection; tracking and classification for pedestrians, vehicles and other predefined object classes based on size and speed of movement; and area control by event triggering when people or objects enter, leave or stay in restricted areas.