Frequently asked questions

Scylla connects to your existing security camera system. At the core of Scylla, there are two main threat recognition modules and one smart decision maker. The smart decision maker - Charon works on top of the two main threat recognition modules to make a final decision. We run the video frame through our deep learning algorithm, and if a detection is made, Scylla will send an alert with the detected frame to your security team or designated staff.

For cloud deployment, you will need to install a small connector software package that will connect to your cameras and forward the streams to the corresponding cloud-hosted Scylla instances. In case of on-premise installation, Scylla is either delivered pre-installed together with the hardware, or it can be deployed and configured by Scylla Tech Support team once they are given remote access in one of the conventional scenarios (VPN is preferable). In either case, you will be provided detailed instructions as well as full guidance from Scylla Technical Support in case if you have any questions.

Scylla can be used in any organization to help increase physical security by providing intelligent and human error-free monitoring and detection of hazardous objects, guns and behavior.

Scylla runs on dedicated GPU-equipped servers. Other than that - it integrates with your existing security camera system, so no additional hardware is required.

The average mass shooting lasts 12.5 minutes, yet it takes law enforcement an average of 18 minutes to neutralize the threat. Scylla AI provides real-time intelligence to first responders on where the active shooter is, what he’s armed and equipped with, what he looks like, and how many shooters there are. This information allows law enforcement to locate and neutralize the threat rapidly and for emergency medical personnel to render aid to the wounded if needed. The instant and informative notification provided by Scylla AI increases the probability of intervention before shots are fired.

Our packages depend on several variables, including the size of the deal, models chosen, deployment, and the additional configuration required. Please contact Scylla or our official partners for pricing.

The accuracy of each solution is carefully and meticulously measured and reported in the official Scylla whitepaper. Moreover, based on AI technologies, Scylla models and algorithms also evolve and improve. Thus the reports are periodically repeated and updated.

Being agnostic to moving backgrounds and having proprietary zoom-in tracking features incorporated in the detection algorithm, Scylla effectively runs on drones and portable cameras.

Scylla modules are agnostic to camera makes and models. The solutions have their minimum requirements, of course. For instance, in most cases, your cameras should support digital output in one of the conventional streaming protocols (RTSP/RTMP/HTTP), have HD resolution, and 5-10 Frames per second imaging speed. Please refer to the minimum requirements provided with the module you are interested in.

Scylla is constantly updating the list of Video Management Systems with which Scylla solutions has one- and two-way integration as integrations are added. At the moment, Scylla works with Milestone, Genetec, Network Optix, Exacqvision, HikCentral, Eagle Eye, and Panasonic, to name a few.

The models that are currently used in production include detection of fight/violent behavior, smoke and fire, and suspicious behavior that can result in shoplifting. Stay tuned for new use cases to be released soon.

Scylla does not store any data unless you decide to do so. Even if you switch on the data storage, Scylla will still store the data on your local network, so you are entirely in charge of your data.

Scylla uses a proprietary Face Recognition model that outperforms state-of-the-art alternatives when tested against industry standard benchmarks. With a “tracking and multiple assessment per person” approach, Scylla provides industry-standard-level person identification accuracy.