How it works
Scylla Face Recognition is built on state-of-the-art deep learning techniques for higher accuracy and identification speed.
Each person entering the view of the camera is being tracked and identified multiple times. The results are statistically verified and the final decision is based on the most reliable output.
Face recognition module is able to work in unconstrained environments, extreme distances and various face angles, requiring as little as one good image of a person for accurate real-time identification.
It allows to minimize biases as the facial recognition system was developed on a dataset balanced with all ethnicities and genders.
It is easily integrated with Access Control Systems to grant entrу based on biometrics or to alert in case a person from the watchlist has been detected.
What makes Scylla Face Recognition Technology stand out
Speed, accuracy, and easiness of deployment through our web and desktop applications.
It can be easily integrated with other modules Scylla provides.
Scylla applies best modern practices in deep learning research to keep the the face recognition system aligned with the accuracy and speed breakthroughs.
To benchmark and compare the accuracy of our algorithms we performed a few most common benchmarking tests. The results are presented in the table below.
Scylla Face Recognition modules operates on edge devices from Nvidia Jetson family (Nano, Xavier, etc.), and the box could be installed for role based access control or other similar use cases.