printer friendly version

False alarm filtering

Off-site monitoring has experienced tremendous growth in the past few years as video surveillance systems have become more widely adopted and cost effective. As the demand for video monitoring grows, so does the need for operational efficiencies to enable businesses to scale.

While the live monitoring of cameras has its benefits, it is not a scalable operational model. ‘Black Screen’ monitoring or event-based monitoring provides some level of filtering, only alerting when an event is triggered. However, even these triggers, usually motion detection or line crossing events, or edge-based AI, are still numerous, and a further level of verification or filtering is required to achieve a manageable number of alarms coming into the control room.

Adverse weather conditions, like rain and wind causing changes in lighting and moving foliage, as well as spider webs and wildlife/pets in the field of view all add to the alarm burden for the monitoring operators.

Nuisance or false alarms are debilitating in the control room, both in the immediate and long term. When operators are overwhelmed with alarms, genuine alarms are not addressed timeously and this causes real security threats to be missed. In the long term, operators become complacent regarding alarms and eventually demoralised as they complete meaningless work addressing nuisance alerts.

At scale, a false alarm filtering system provides significant benefits for a monitoring operation.

False alarm filtering benefits

A false alarm filtering system receives the data (image) from an initial event (e.g. line crossing) and analyses the image to determine if it was a person or vehicle causing the event. An event verified as true, such as a person or vehicle detected, is then delivered to the alarm stack while false events are filtered out and ignored.

There are various options for monitoring companies to deploy a false alarm filtering system, the detection capabilities on the camera/NVR, a bespoke server on site or, a cloud-based solution. Each has its own pros and cons, however a cloud-based solution is by far the most powerful.

While the name ‘false alarm filtering service’ suggest that filtering out false alarms is the primary function (which it is), this function actually rests entirely on the ability of the system to accurately detect people and/or vehicles under various conditions. Distance from the camera, low lighting, poor contrast, unusual camera angles and posture of object (e.g. crawling), fast moving objects and poor resolution of image all impact the ability to accurately detect objects.

The greater computing power of cloud-based solutions and the ability of these systems to be iteratively improved by constantly increasing the dataset from which the system learns, means that the true power of AI is unleashed, and this delivers a higher level of detection accuracy.

So, in a perverse way, the false alarm filtering system is actually the gatekeeper of the control room, and its first job is not to miss any true alarms. When viewed in that light, it is understandably critical that any false alarm filtering service is, in the first instance, accurate in its detections.

However, even the computing power of cloud-based solutions using AI models are not infallible. This is where the configurability of a system with respect to sensitivities and thresholds is important. There is an inherent trade-off between filtering out more nuisance alarms and possibly missing a genuine alarm. The ability to configure a system and find the ‘sweet spot’ is vital.

Focus on the real issues

Monitoring operations wanting to scale are faced with the choice of employing more operators or deploying an effective false alarm filtering system. A well configured system will ensure that there is little chance of missed detections and a reduction of about 80% of nuisance alarms (compared to motion detection or line crossing on cameras).

When operators are addressing genuine threats without the distraction of constant nuisance alarms, they are able to focus more clearly on the genuine threats, address them timeously and deliver a better service. There is less complacency and morale is improved as most of the work done is meaningful responses to genuine alarms.

Many leading South African off-site monitoring companies are already making use of a false alarm filtering service to augment their operations and keep their operational costs under control.

There is no doubt that if your plan is to scale your monitoring operations, a false alarm filtering service is critical, rather than a nice-to-have. As the technology develops, accuracy improves and pricing become more affordable. A consideration and assessment of a false alarm filtering service for your monitoring operation is certainly well worth the effort.

For more information, contact DeepAlert, +27 21 201 7111, sales@deepalert.ai, www.deepalert.ai

Share this article:

Further reading: