AI.SECURITY SOLUTIONS: Monitoring of fencing systems using structure-borne sound and artificial intelligence (AI). The system is based on the patented Truck Norris system.
NO CHANCE FOR BURGLARS
By means of piezo sensors, structure-borne sound is registered even over long distances in the fence meshes. An artificial intelligence trained on fence cuts analyzes the sounds. Background noises are detected as well as the attempt to climb the fence. A multi-stage alarm system differentiates between burglary, damage caused by animals or attempts at tampering.
PATENTED AUDIO SYSTEM
The Cut-Spencer system is based on the same technology as the proven and patented Truck Norris system, which is successfully used against truck tarpaulin slashers. The innovative Audio Recognition Framework, an AI pipeline, has been trained on a variety of hazardous sounds. It is 100% GDPR compliant and can be set up on the European cloud, as well as on-premise on your own server.
A small compute unit analyzes noises within the wire, which are recognized by a structure-borne sound microphone.
Alarms from the small compute units are subjected to more accurate ML analysis in the Central-Unit to avoid any false alarms.
Even miles of wire mesh fencing can be monitored in 500-meter segments without gaps. The Recognizer modules (embedded microprocessor systems) are connected either via Ethernet, GSM or WLAN.
A wide range of connection protocols also enables use in regions with low network coverage (GSM/WLAN/NB-IoT/LoraWAN). Because of the versatile plug-and-play combinations with other sensors that can be used, our systems can be integrated into various SMART CITY and SECURITY SOLUTIONS.
AI, Audio and IoT: Alarm Systems with Innovative Control
Piezo sensors are mounted on the fence
The small compute units (Smart Dust) activate an alarm signal as soon as the set threshold is exceeded by a suspicious sound
The TN Central-Unit (CU) uses AI to analyze the transmitted sound.
Can the alarm trigger be integrated with a video management system?
Yes, if the integration consists of starting and stopping the video recording. Integration with existing security systems is possible via our API. Activation of additional cameras in public spaces after an alarm is triggered does not violate privacy laws.
What are the components? What is the power consumption?
Cut Spencer uses piezo structure-borne sound sensors on Arduino-like systems. The sensor detects vibrations, bends, shocks and touches on the fence even over long distances.
How accurately can the trigger be localized for 200 meter sections?
The integrated AI sensor already has a very high accuracy. Nevertheless, suspicious sounds are subjected to a further check by a second, more powerful AI before an alarm is triggered. This prevents so-called “false positives”.
How many systems are needed to secure fences?
This depends largely on the structure of the fence. If a metal fence is completely connected, i.e. does not consist of several interrupted segments, fewer microphones (sensors) are required. After an initial test on site, the optimal number of sensors and devices is determined. One Central-Unit takes care of the complete array of serial smart sensors.
Ready to use in just fifteen minutes for years of peace of mind.
Our systems are designed to help recognition software become far more aware of what is happening in front of and behind the fence. The alarm system learns – and we learn with it.
Unusual sounds are actively learned by the system
Powerplay at the Sparkassen-Eisdom in Halle: With the digital solution from Infinite Devices, the Sparkassen-Eisdom is once again open for public skating according to the 3-G rule.
How can you safely measure visitor flows in pandemic times? We have developed the innovative “SmartCountr” access control system for this purpose, which is already being used successfully by many customers in Germany.
“To date, little attention has been paid by the research community to the applications of internetofthings (IoT) in the field of interactive sonification to the shared control of the sound system by both the user performing gestures local to the system itself and by one or more remote users.”