Get foreground mask without updating model
I want to write an algorithm that detects camera tampering e.g blocking the view with piece of paper etc, so I thought about measuring the differences between current and background frame. I need to measure the level of these differences before updating MOG2 model then update the model. Can I do that?
I mean, that I don't want to update model with completely different image that is a result of tampering like piece of paper with random drawing in front of camera. Also I want to have only one MOG2 instance in the system (performance reasons).
why (ab)use Background Subtractor classes, when it's clearly not what you want or need ?