Delay in CamPlus AI notifications

What delays in notification are people getting from the time a person or vehicle enters the camera’s Field of Vision (FoV) to the time a notification is generated? I’m typically getting between 14 and 22 seconds, which is frankly too much of a delay to be useful, aside from forensics review.

It might be helpful to have either a camera with sufficient processing power to do the edge processing, or software on a local micro-server like an RPi 4 perform the model inferencing on the local network.