E:T utilizes world-class A.I(Artificial Intelligence) technology called DNN(Deep Neural Networks) or Deep Learning. E:T applies DNN computer vision algorithms or deep-learning methods called object-detection on the image frames of your real-time security camera footage. It then processes the results based upon a computer vision A.I model. Detection results are then compared with the objects you defined within the list of object classes available from the A.I model.

For example:

A.I models is usually pre-trained to detect only certain objects classes like persons, dogs, cats, cars and buses. If you choose this specific model, then you tell E:T to notify you on whether anyone of your camera feeds detects any one of the objects defined in the models class list. You could also choose specific objects like ‘person’ or a ‘dog’ and E:T will notify you on exactly that. It is a form of digital intelligence that aims to replicate the cognitive abilities the same as that of a human being.

Why is this important?

Most standard security CCTV systems deploy pixel-based methodologies and alarms on events such as motion. These motion-based triggers can be caused by anything moving in front of the camera. From leaves blown by the wind to flashing lights to insects moving across the lens, these are all things you don’t care about and these are things which mostly yield false alarms being triggered and annoy you through notifications which you’ll eventually switch off.

Yes, some standard CCTV systems depending on the price range have built-in intelligence like facial detection and people counting etcetera, but that is how far it goes for the cheaper systems. What if you want to know when certain things appear in front of your camera? Things like a car, or a motorbike or a sheep or a cow? These are just some of the endless things which could be detected through DNN or deep-learning provided its model is trained on those objects. The world of Deep Learning has opened the doors to digital intelligence and E:T deploys this kind of technology in conjunction with other technologies to solve many of the surveillance challenges.

E:T was not designed to replace conventional CCTV systems, but to complement and enhance an existing system. It can be viewed as an intelligence plug-in into your existing security ecosystem. And E:T, in turn, provides you with a powerful capability to act real-time from anywhere in the world! You don’t have to replace your existing CCTV system, just plug-in E:T.

How does E:T “plug-in”?

Raspberry Pi

First of all, E:T can work on any computer provided that the system environmental requirements are met. However E:T was designed around cheap SBC’s(Single Board Computers) available anywhere around the world, specifically the beloved Raspberry Pi together with Intel NCS2.

The Pi as it is referred to is a stand-alone computer that requires nothing else except external power. The Pi can work totally headless: meaning no screen, mouse or keyboard is required for it to function as a normal PC. This makes it an ideal computing platform for E:T to function. The Raspberry Pi has everything it needs to communicate with the outside world. Best of all, the Pi is:

  • Small
  • Affordable
  • Reliable
  • and, widely supported by makers and enthusiasts around the world

Although the Pi has well-respected performance compared to its form factor, it is however not suited for Deep Learning A.I performance. Or at least not for E:T’s multi-stream real-time requirements. Thus with the release of Intel NCS2 the Pi received A.I DNN capability and thus served as the ideal hardware platform for E:T.

So simply connect the NCS2 to the Pi’s USB port, setup the E:T software, connect the Pi to your network and E:T is now “plugged-in”, ready to stream and be your digital agent. Below are before/after diagrams of a typical CCTV setup without E:T and then an after with E:T added:

Before

After