We are all fascinated by the ability of AI to describe what is happening in front of a video camera, but so far we have not seen anyone answer the question — why? Is there at least one truly practical use case?
At SpecLab, however, we came up with our own application.
A familiar picture, isn’t it?
Of course, you could wake up a security guard by using the same neural networks to detect when he falls asleep. But if you force him to stay constantly alert, he will simply quit (especially considering the salary security guards are usually paid).
It is much more effective to gently draw his attention to what is happening.
We integrated a NEURAL PROMPTER into our event-based video surveillance system. It verbally describes events appearing on the screen, for example:
A car is leaving the parking lot
A person walked past the house
People are walking down the corridor
An unfamiliar person has been standing near the entrance for a long time
People wearing masks entered the warehouse
A pack of dogs has been detected near the exit of the building
Two cars approached the house
A truck with an empty cargo compartment is passing through the checkpoint
A truck fully loaded with sand drove down the road
A person fell in the crane operating area
A drone flew past the north entrance
A crowd of people is standing near the checkpoint entrance
Armed individuals detected inside the warehouse
Attention! A person has fallen. Please assist them
Naturally, it cannot speak constantly like this, otherwise it would quickly become annoying. The neural analytics system can be configured with a minimal interval during which it selects only the most important message, for example:
“A car is leaving the parking lot, but nobody has exited the office…”
A small but meaningful victory over the human factor.
It is also nice that the neural networks are fully integrated into SpecLab’s video surveillance software and do not require an Internet connection.
Olesya Grishanina
=SpecLab=
