Page 16 - Security Today, March 2021

Transforming the Industry AI enabled cameras are poised to make a significant impact in security
Artificial Intelligence (AI) con- tinues to gain momentum every day, and it is already poised to augment and enrich many as- pects of our business and personal lives. For the security industry, and for video surveil- lance in particular, it’s clear that AI-based technology is about to make a substantial impact. Let’s examine some basic questions: How does it really work? What can it do for business? Is it going to be painful to migrate from non-AI cameras to AI-enabled ones?
According to a recent IHS Market Video Surveillance Installed Base Report, close to 85 million cameras will be installed in North America alone by 2021. But we cannot expect a proportional increase in se- curity personnel to monitor and manually search through this vast amount of video. AI presents a perfect solution to compen- sate for unmanned environments or those with limited staffing, or the loss of vigi- lance after looking at a screen too long. AI can help us not only watch continuously, but also feed systems that are able to sort, organize and categorize massive amounts of data in a way that human operators can- not. And it can do so far more reliably than traditional video analytics ever did.
Dispelling Myths
and Misconceptions
Correctly teaching an AI algorithm and ensuring its accuracy is done in a sophis- ticated process, even for server-based AI models and solutions. It is even more chal- lenging to optimize deep learning mod- els on the camera edge. Unlike a server implementation with far greater compute power, storage and database resources, or a cloud-based system with significant scal- ability, AI deployed in edge devices such as cameras have limited computational power to identify and classify objects.
Once the algorithms have been pre- trained to identify certain objects and characteristics, they do not have the capa- bility to “learn new things” by themselves. Additional capability is deployed by re- peating the above process and deploying new firmware to the camera and its deep learning accelerator resources. This al-
lows for a lighter approach to AI hardware resources on the camera, and still allows strong capability to deploy to the edge which can get stronger as it evolves.
A deep learning algorithm doesn’t see video in the same way we do, but it can
look for familiar shapes and patterns that it has been trained to recognize. It can’t think for itself or make decisions that it hasn’t specifically been programmed to make. AI-based cameras aren’t inherently doing anything a human can’t do, but they
16
MARCH 2021 | SECURITY TODAY
COVER STORY