[logophobia]

Consider using background subtraction for this: https://docs.opencv.org/4.x/d1/dc5/tutorial_background_subtr...

Will make things a little more robust (and overengineered!).

[catlifeonmars]

I’m curious, could you somehow make the background model adaptive by averaging a rolling window of camera frames to use as the background model in this?

[maxmunzel]

This is pretty much what I’m doing. Except that I’m using an exponential moving average so I don’t have to mess with windowing.

[catlifeonmars]

Oh nice, I hadn’t thought of that.

[maxmunzel]

Thanks for the hint!