Deep learning AI is fast becoming crucial within the Security Industry. It has the potential to make systems more automated by requiring fewer human interventions. Deep Learning Artificial Intelligence can be utilised to great effect when establishing a long-term solution for Video Analytics. It could make Security Solutions significantly more efficient over the next few years, so here’s our rundown of the core elements of a deep learning AI box.  
 

The Concept of Deep Learning

Deep learning radically advances false alarm filtering in both Residential Security as well as Commercial Security. It aims to notify facilities with just the events that matter by identifying people and vehicles more reliably and more precisely. This provides a vastly more effective approach to filtering out false alarms when compared to motion detection and traditional video analytics.
 

How does a Deep Learning AI Box work

Because it utilises multi-layer networks of algorithms, rather than simply looking for changes in images, a deep learning AI box can better learn what an object is.

Through combining pixel values, edges, vector shapes and a host of other elements, Johnson Controls AI products have been designed to provide comprehensive security. By Imitating the way a human gains and adapts to new knowledge, it can effectively learn how a person or a vehicle moves and what they look like. Furthermore, it can identify this from a range of different angles, distances, and lighting conditions.
 

Key Features

The core elements of the deep leaning AI box include:

• Object detection and analysis
• Event triggers (including establishing a perimeter defence, out of hours detection, and no parking identification zones)
• Built-in rules: human and vehicle detection, illegal behaviour, and an object counting system
    

Reducing false alarms in video analytics

False alarms cost a facility time and money, whilst also wasting the resources of emergency services and causing reputational damage. Thus, deep learning AI aims to reduce false alarms through:

• Static object filters
• Confidence thresholds for every detected object
• A double knock feature designed to respond only if both the deep learning and the PIR sensor trigger an event simultaneously
• Ignore areas and dead zones
    

Video Surveillance by Johnson Controls

At Johnson Controls, we supply built-in deep learning as well as AI-based video analytics as part of our integrated security solutions. Our systems offer extras such as Deep Learning Plus, which has an exciting, immensely powerful expansion device in either 4, 8 or 16 channel options.

For more information about how Johnson Controls can ensure the safety and security of your premises, check out our Video Surveillance Products Here. You can also learn more about how Deep Learning and in particular Black Box AI are revolutionising business innovation and efficiency here.