The Market

The surveillance industry is served, for the most part, by companies that address specific niche markets and thus cater to a small part of the total addressable market.

Artificial intelligence (AI) is widely used in the surveillance security market, but machine learning has not yet been made adaptive - that is, until now.

So far, AI code has been written for specific abnormal behaviors, but programmers must know the behavior they are looking for, or the system will not recognize it. Because of this, separate video-surveillance products need to be produced for each situation or behavior, which is cost prohibitive, labor intensive, inflexible and difficult to implement. This approach also leads to one of the most time-consuming and distracting phenomenon in the industry: false positives. A false positive is created when image analytics software reports normal behavior as seemingly suspicious, overloading the operator with non-relevant information, thereby occupying valuable resources and compromising the chances of capturing a true security breach."

In addition, existing technologies have other significant problems with their perception engines: resolution is so poor the human form cannot always be detected; images are heavily distorted by environmental conditions; and the human form can blend into backgrounds, making it undetectable. Existing recognition software is also generally limited to forensic analysis - but after the event, it is too late.

Another industry challenge is the need for software that can work with legacy video cameras, avoiding the Herculean effort of upgrading the infrastructure.