Unusual Behaviour

Pixel-based analysis allows the system to learn how objects normally move in an environment taking into account direction, speed, size and a variety of other factors. After a norm is established for a particular scene, the system will then create an event on any exceptions. These events are then classified using Deep Learning tools and a logic engine for further contextualisation.

iSentry Unusual Behaviour is designed to add instant visual insight (much like a human eye) to the huge amounts of footage generated by a large scale CCTV deployment.

Threat Detection and Extraction

Threat Detection and Extraction (TREX) module, utilises a pixel-based dynamic AI learning process to acquire and track objects of interest in real-time. To some extent, TREX mimics human vision where we pay more attention to sudden movements and pay less attention to regular movement. As such TREX has the capability to learn the scene to focus on true targets and not on environmental factors that are common to the scene.

Important in this process is to ignore environmental factors inherent to real-world cameras, such moving trees, water reflections etc.. TREX is capable of tolerating a wide variety of camera noise. TREX is able to detect fast moving or very small targets, which can translate to targets at long distance away.

Special Applications

Customised security, safety and compliance solutions. The power of iSentry is complemented by a variety of additional customisable tools, enriching the outcome of the UB and TREX processes and leveraging the power of our deep learning and rules engine modules. E.g.:

• Face Detection & Recognition
• Cash Detection
• Helmet (compliance) Detection

COVID-19

Key aspects across the board in all de-escalation plans in all major countries across the world will almost certainly involve:

• “social distancing”, i.e. keeping a certain distance between persons
• personal protection, i.e. face masks
• capacity and crowd control