Perimeter Intrusion Detection System
Scylla high-precision AI surveillance technology can effectively filter out up to 99.95% of false alarms, minimizing noise fatigue, unnecessary panic and expenses.
Book a demoRelated materials
FAQ
Scylla Intrusion Detection and Perimeter Protection System is one of the most flexible solutions as it can be easily integrated with most VMS and CCTV systems. Depending on the infrastructure the client already has on-site, the integration can be implemented one- or two-way. In case of one-way integration, Scylla IDS uses video streams from existing cameras directly or through NVR/DVR they are connected to. The analysis output, in this case, is shown in Scylla's dashboard. In case of two-way integration (e.g. with major VMS providers, such as Mobotix, Genetec, NX Witness, etc.), the output can be viewed in the corresponding VMS dashboard.
Typically less than a second. In cloud deployments, the response time can slightly increase depending on the client’s upload speed.
Currently, the objects detected by Scylla Perimeter Intrusion Detection System include humans and/or several types of vehicles (car, truck, bus, van, motorcycle).
Absolutely. Scylla does not store any data that can be considered personal. No footage or images are stored. The only cases when we store data is the reported alerts. Their retention time can be adjusted to comply with regulations the client has in place.
Yes, it does. Moreover, almost all Scylla products can be deployed on cloud. In the cloud-hosted scenario, the video/frames should be forwarded from CCTV cameras to the cloud in one of the following ways:
1) by using the proprietary Scylla Connector software that connects to the camera, takes the video stream, encodes it, and sends it packet-by-packet to the cloud, where it is further decoded and analyzed. This architecture is favored where there are some restrictions (e.g. a domain, port, etc.) in the local network.
2) by utilizing the embedded algorithms of the camera (i.e., motion detection) that send one or several frames, or a video chunk to the cloud through HTTP/FTP and similar protocols. This scenario is favored when the client is looking to minimize their upload bandwidth. This flow is implemented in Scylla False Alarm Filtering System (see further).
3) by exposing cameras to the cloud directly. This approach is the least popular due to related technical difficulties and security concerns.
An alert is classified as a true alarm when the prediction of AI corresponds with the reality (i.e. the object of interest is correctly identified, the action sought after is detected, etc.). A false positive is a case when the alert is triggered by mistake. Unfortunately, due to the essentially probabilistic nature of AI, the latter are inevitable in most cases. However, due to the elaborate AI and machine learning behind Scylla Perimeter Intrusion Detection System, as well as embedded patented algorithms such as "eagle eye" it can meet any level of production-grade industrial standards. Moreover, we are continuously improving Scylla AI video analytics modules where they are retrained on mistakes to make sure the number of false alarms further decreases with time.
The duration of alerts depends on the client's data retention policy. By default, we offer a storage duration of one month, but this period can be configured to correspond to local policies.
Yes, Scylla Intrusion Detection System easily analyzes video feeds from PTZ cameras.
Scylla Perimeter Intrusion Detection technology is essentially camera-agnostic. Most questions on the limitations and camera requirements end up receiving a simplified “rule-of-thumb” answer — if a human can see and identify the object of interest, then Scylla Intruder Detection AI will also be able to do that (and in some cases will even outperform due to the integrated zooming and re-checking algorithms).
As for the minimal camera parameters, these will depend on your use case and the object of interest you are trying to detect. Of course, the camera should have a digital output or at least be connected to a DVR that has one. Scylla Intrusion Detection and Perimeter Protection System can accept pretty much all the variety of stream types, such as RTSP/RTMP, HTTP, etc. Usually, the minimum required resolution starts from VGA (640x480) and 10 FPS. Parameters defining the frame/image quality vary from one camera to another, but we recommend looking into such characteristics as bandwidth, encoding, and sharpness, and improving them if necessary. Lastly, if Cloud+FTP implementation is desired, the camera should have a sensitive motion detection function with a possibility of uploading to an FTP server of preferably more than one frame.
Scylla Intrusion Detection and Perimeter Protection System works pretty much the way human vision does — not only does it see an “odd” object in the scene, but it also classifies its origin. Scylla AI video analytics is trained on a huge and versatile dataset, therefore, it can effectively recognize humans or different vehicle types. The alarm is formed only if Scylla AI detects one of these objects or parts thereof.
Scylla Intrusion Detection and Perimeter Protection System alerts the corresponding security unit/individual about the act of intrusion. The alert contains invaluable visual and meta-information about the origin of the event of intrusion. Moreover, the intruder detection system can be connected to ACS (Access Control Systems) to lock infrastructure, alert and disable entry protocols, etc.
The limitations here highly depend on the camera specifications (see above). The most important factor is the resolution and view angle of the camera which eventually results in the pixel size of the object in the frame streamed by the camera. Similarly, factors such as illumination, capture contrast, and video streaming parameters (bandwidth/encoding) are to be considered. As a rule of thumb, the distance can be defined as following: if a human can look at the footage and reliably tell that it’s the object of interest (human/vehicle) within the region of interest, then Scylla AI should also be able to do so. Scylla Perimeter Intrusion Detection System installed on 4K drone cameras can detect humans at distances of up to ~200m. Alternatively, one can use the following information to define the limitations: the minimum height of a person on the frame should be ~30 pixels, and that’s ~1% of 4K images.
The notification can reach the end-user through one of the alerting pathways, namely:
1) Scylla’s dedicated web-based dashboard that shows alerts and allows to configure the system and monitor your environment,
2) Mobile alerting application, Client’s VMS dashboard (in case it is supported by Scylla Perimeter Intrusion Detection technology, e.g. Milestone, Genetec, NX Witness, etc.),
3) Access Control System notifications, i.e. signal lights and sounds attached to alerting relay boards.
Scylla Perimeter Intrusion Detection is designed to work in challenging environments where cameras with embedded algorithms do not perform. The AI engine compensates for the drawbacks imposed by demanding conditions including but not limited to poor illumination, somewhat corrupted frames, environmental factors, and weather-foisted effects.
Yes, it can, including border control distance IR/thermal cameras. The DRI parameters (Detection, Recognition and Identification) of Scylla Perimeter Intrusion Detection will depend on the camera characteristics (contrast ratio, pixel crosstalk, etc.). But in general, the solution complies with industry-standard DRI requirements, i.e. the recognition limit (the distance at which you can determine an object’s class - is it a human or a car, a truck or a tank, etc.) is ~15 pixels for humans and vehicles.
If Scylla Facial Recognition Module is deployed together with Intrusion Detection and Perimeter Protection System, it can enable white- and watchlisting of individuals detected on site. However, as Face Recognition module is a standalone solution that is optional for Scylla Perimeter Intrusion Detection System, a few points are to be considered:
1) To run Facial Recognition Module you have to provide additional hardware. Face recognition requires additional ~2Gb GPU memory and some computing capacity.
2) Accuracy of Face Recognition Module in “watchlist implementation” highly depends on a number of external factors, such as the visibility of the face of the individual, the size, the angle of it, possible obscurations, etc. While Intrusion Detection can be triggered when even a small part of a body is in the active area, Face Recognition is more demanding. To boost the accuracy of Facial Recognition in that respect Scylla Perimeter Intrusion Detection needs to be implemented together with Person Tracking to increase the chances of identification in cases when the individual’s biometrics are visible. However, Person Tracking is a resource-demanding algorithm that will require additional hardware.
3) As mentioned above - Face Recognition Module is a separate product that needs to be purchased separately.
PPFs (pixels per foot, a parameter that defines the maximum distance for detection) are 10 for humans, and 5 for cars. Note that in general these parameters are also a function of other conditions such as camera quality (contrast, SNR..), illumination, etc. So they are to be treated as statistically valid references.
In short – Scylla False Alarm Filtering is a subsidiary module of Scylla Intrusion Detection. Both solutions have the same goal – detection of intrusion into defined zones. Both solutions are based essentially on the same AI models and logic. However, Scylla False Alarm Filtering depends on preliminary embedded analytics of camera/VMS motion detection modules. What happens there is: 1) the camera/VMS detects a motion in a predefined zone, 2) the camera sends a chunk of a video (preferably) or an image to Scylla False Alarm Filtering for analysis 3) Scylla False Alarm Filtering checks if the movement is caused by a person or vehicle, 4) in case the movement is verified a corresponding alert is formed and sent to end-user.
In contrast, Scylla Perimeter Intrusion Detection system takes care of the whole sequence including steps 1) and 2).
Scylla False Alarm Filtering is specifically designed for monitoring centers. It assumes that your setup is already equipped with units capable of motion detection and forwarding of video chunks. Basically, Scylla False Alarm Filtering should be implemented in cases you have a network of intrusion monitoring security cameras and you want to reduce the strain on watching agents from annoying false alerts.
Scylla Intrusion Detection should be used if your network/cameras are not equipped with comprehensive motion detection modules. It can also be advantageous if the embedded alerting module of the VMS/Camera results in deterioration of images.