A.I. Teach’s A.I.-Intrusion is designed to detect intrusions in real-time using the latest A.I. algorithms. It can detect objects such as humans, animals, and vehicles. It can even be configured to recognise specific types of objects. It also provides advanced filtering options, which can help reduce false positives in even the most complex scenarios.
A.I.-Intrusion is a video analytics application by A.I. Tech for intrusion detection. It can work both on optic CCTV and Thermal cameras and provides advanced object filtering and rules to reduce the number of false positives also in very complex scenarios.
A.I. Teach’s A.I.-Intrusion is a finalist in the Benchmark Innovation Awards 2023 in the Analytics and Software Innovation category.
Alessia Saggese, Co-Owner at A.I. Tech, explains, “Most simply, the object detected can be filtered considering their pixel size; however, the solution provides an easy calibration procedure for filtering objects based on their real size in meters, enabling the user to control the perspective in the scene accordingly. One of the most important aspects of a filter is the ability to filter objects based on their aspect. There is, in fact, the possibility of using a classifier based on a deep neural network to confirm that the detected object is not an animal but rather a person. As a result, one can filter out objects that are not humans or vehicles, for example.”
A.I.-Intrusion allows the user to configure different types of virtual sensors:
- Area sensor or loitering: for monitoring the persistence of intrusions or loitering within an area of interest framed by the camera. A sensor can be configured to notify of intrusions or loitering alarms if an object of interest remains within the area beyond a specified minimum period.
- Single line sensor: is a sensor used to detect the intrusion of a person who crosses a virtual border determined in advance within a predetermined area.
- Multiple line sensor: is used as a tool for detecting the intrusion of a person crossing several virtual lines (not necessarily in parallel) for a certain period of time.
A.I.-Intrusion also allows managing two configurations that can be automatically switched using a timetable. This functionality enables a user to configure different sensors or filtering at different times of the day; for instance, a user may have a configuration that is active during the day and the other for the night.
Two new features are available:
- Visualisation of intruders on a virtual map. This advanced view allows a human operator to track the movement of intruders over a map combining the information coming from different video streams processed by AI-Intrusion. No matter how many cameras are under analysis. (See an example of the video below).
- GPS coordinates of intruders. The application can be configured to provide, as additional metadata, the GPS position of objects involved in alarms.
Does A.I. intrusion help to save money or resources?
“Different customers can get benefits using A.I. Intrusion,” comments Alessia and lists:
- Adding video analytics capabilities to preinstalled cameras can help homeowners improve their home security system by adding them as additional sensors to their preinstalled cameras. The alarms detected by A.I. Intrusion can also be transmitted through a network or digital I/O to the anti-theft unit. This is especially useful for people with gardens or open spaces where it is complex or expensive to have total coverage using traditional sensors.
- A.I. Intrusion is very beneficial for companies that require a system to monitor perimeters and forbidden areas of industrial buildings automatically, day and night. Without this solution, they would need people to patrol these areas and monitor them both remotely and in person. The solution may reduce the number of people required and avoid exposing them directly to potential threats.
- A surveillance company may use this solution in conjunction with its VMS and PSIM to help operators while monitoring many cameras connected to them. In this way, the latter may pay attention only to those cameras for which the application has notified an alarm. Moreover, since fewer people are required to monitor the same site, the company may be able to scale its business and monitor more sites.
A.I.-intrusion UX
A.I.-Intrusion offers the user an easy-to-use web interface that makes it easy to set up sensors, calibrate the app, and check the configuration right on the live stream to provide immediate feedback on the configuration changes.
A new version of A.I.-Intrusion A.I.-Intrusion is available as a ready-to-install installation package for a range of surveillance cameras manufactured by Hanwha Techwin, i-Pro, and Axis, as well as already being available for cameras of the 7th series from Mobotix.
Packages are also available that can be used on servers and embedded devices.
Finally, A.I.-Intrusion is distributed on our AI-App, a ready-to-use hardware box to be installed on a network and provide video analytics capabilities to new and preinstalled cameras.
There are also manuals, tutorials and academies available on the A.I. Tech web portal ( https://help.aitech.vision ) for users to learn how to configure the solution and get assistance on how to set it up.
To vote for A.I. Intrusion, click here