I'm building a security project on a Raspberry Pi and want to ask the following:
Is motion detection and facial recognition best done in real time or taken from still photo?
I want to build a camera that has motion detection but can also recognize if there are any humans in the image, and only send an alert if that motion is proved to be human. Originally, I was going to take some images, average them out and then check, but I've seen some algorithms that work from a video feed.
I know my question is too broad in its current form. Is there any way of rewording it so it's less so. Or is it best to not ask it altogether.
Being a relatively low-rep user on the site I thought it would be better to ask here than to ask the question and see.
I've gone ahead and asked this question. However, another problem that is coming up is that it's too software based. In which case, could it be migrated to Stack Overflow?