Face Recognition

Upgrade your video surveillance with Scylla Face Recognition that harnesses the power of computer vision and artificial intelligence for the safety of your premises and people.

Book a demo

Related materials

How to ensure accuracy and sensitivity of the face recognition system

Face recognition technology is widely used by security experts worldwide. Armen Ghambaryan, Ph.D. shares about the stages required to ensure that the facial recognition system operates accurately and effectively.

Face Recognition Technology: How it Works

Face recognition technology is expected to generate $9.6 billion dollars by 2022. Explore how facial recognition systems work and what causes such a fast-paced growth for this industry.

How AI Video Analytics can Improve School Security

Over the past decade, 180 school shooting incidents have been reported in the US. It is time to embrace intelligent school security systems that leverage the power of AI and machine learning to detect and report anomalies and threats in real time.

FAQ

Face Recognition accuracy is indeed impaired by partial obscuration. However, in good setup conditions (camera, illumination, proximity, etc.) Scylla Face Recognition is still capable of accurate Face Recognition (shown in a number of tests).

At least one picture of a person is needed for the system to recognize a face. 45° seems to be the golden middle for the angle.

Scylla Face recognition can detect a person from just one frame. Minimal data is required for a positive match. The system is trained on all possible angles of objects of interest, it works in difficult camera environments (extreme angles and distances). Of course, in some specific cases, the angle would matter as the features that Scylla uses to recognize a face are more distinct in some angles than others. Facial detection can be done from the side of a face – up to 90°, before accuracy gets poor.

Scylla Face Recognition has a wide variety of use cases and can be used for missing people detection. Combined with Scylla Person Search System that utilizes AI and computer vision for appearance-based search, the Face Recognition module can help increase the speed and chances of finding a missing person across the connected CCTV cameras, forensically or in real-time.

The training set minimizes detection failures with people of diverse backgrounds. There have been reports of some face recognition models not working well with people of color and women due to imbalanced data sets used for training. Many facial recognition systems are trained on images of white males only. Scylla has deliberately built ethnically and gender balanced datasets in order to eliminate bias in face recognition.

Scylla does not store any data that can be considered personal. No footage or images are stored. In jurisdictions where face recognition is not permitted for privacy reasons, Scylla’s Person Search can be used to search for someone based on their general appearance.

Face Recognition can be seamlessly integrated with most Access Control Systems and extend the capabilities of other Scylla's solutions.

Scylla's Face Recognition module works on edge devices from Nvidia Jetson family (Nano, Xavier, etc.), and Asteria edge device can be effectively used for access control monitoring or other similar use cases.

The number of faces analyzed per stream is limited only with hardware capacities. We do not store faces in the system, only the face vectors which are one-directional encryptions, hence it is not possible to reverse engineer them.

Face Recognition can be used for many use cases, such as: ticket-free biometrics check-in, BOLO detection, VIP customer detection, employee access, etc.

A person can still be detected in a broad variety of appearance and age changes.

Face Recognition

Related materials

How to ensure accuracy and sensitivity of the face recognition system

Face Recognition Technology: How it Works

How AI Video Analytics can Improve School Security

FAQ

How many pictures do we need for Face Recognition? What is the optimum camera tilt angle your software expects for best detection?

Can the person of interest be at an angle and the system still can identify him/her correctly?

Can Face Recognition be used to search missing people?

How is Scylla ethically approaching Face Recognition?

How personal data are collected and stored? How can the privacy of data be protected?

Can Face Recognition be integrated with access control solutions or other Scylla's solutions?

Does this solution work on edge devices?

How many faces can the Face Recognition detect at one time?

What purposes can Face Recognition be used for?

Can a person from a watchlist be detected if he/she changes his/her appearance (e.g. if a person puts on glasses, a false beard, a wig, etc)?