Дякуємо, ми отримали ваше повідомлення і звʼяжемось в найближчий час! :)
How it works:
● Scylla’s facial recognition watchlist uses deep learning techniques to ensure maximum accuracy and speed in identification.
● Individuals in the camera’s field of view are tracked and identified multiple times, with results statistically validated to determine the most reliable match.
● The module operates in various conditions, at long distances, and from any angle, requiring only one quality image for accurate real-time identification.
● The system minimizes bias through a balanced dataset across ethnic and gender groups.
● It integrates seamlessly with access control systems, either by using biometric data or alerting when watchlisted individuals are detected.
What does Scylla Face Recognition Watchlist detect?
Scylla Face Recognition supports a wide range of use cases, including:
Access control systems
Attendance tracking
Watchlist monitoring
What sets Scylla’s facial recognition technology apart from other surveillance systems?
● Speed, accuracy, and ease of deployment through our web and desktop applications.
● Seamless integration with other Scylla modules, including the Scylla Access Control solution.
● Scylla adopts cutting-edge practices in deep learning research to keep its facial recognition system aligned with global advancements in accuracy and speed.
● The Scylla Face Recognition module works effectively with Asteria monitoring devices for role-based access control and similar use cases.
FAQ
How does Scylla store faces and profiles?
The facial recognition system uses two distinct databases:
– The vector database, which stores irreversible facial vectors — ensuring that the original image cannot be reconstructed from these vectors.
– The standard image database, which stores actual facial images. However, storing images is optional.
At the client’s request, the system can operate exclusively with the vector database, ensuring enhanced privacy.
Can the facial recognition feature identify faces wearing masks? If a person’s face is partially covered, can they still be recognized?
Recognition accuracy does decrease with partial occlusion. However, under optimal setup conditions (camera quality, lighting, proximity, etc.), the system can still accurately identify faces — as demonstrated in multiple tests.
Can we run a pilot program before full deployment? How will success be measured?
Yes, a pilot program can be conducted.
Success is measured by the percentage of successful detections under various conditions, aiming for results as close to 100% as possible. The evaluation is based on visual recognition acceptability criteria.
What is the detection range of the facial recognition system?
For Full HD video streams, 250 PPM (pixels per meter) is required, which typically corresponds to a detection range of about 4 meters.
The system's latency ranges from 3 to 4 seconds, depending on specific configuration settings.
Can a person be re-identified if their appearance changes (e.g., wearing glasses, a fake beard, or a wig)?
Yes, the system is designed to operate effectively even with significant transformations or changes in appearance.
How does Scylla uphold ethical standards in training its facial recognition technology?
Scylla minimizes recognition errors across diverse populations by intentionally creating ethnically and gender-balanced training datasets. This approach helps eliminate bias in facial recognition and ensures fair, accurate identification for all individuals.
Fill out the form to receive a personalized consultation for your IT infrastructure: