At night time my Reolink cameras really accentuate any cobwebs in the IR images.
I’m guessing in this day in age it must be possible to analyze the image at 2am and send an alert at 8am if any cameras need a clean?
Could someone suggest some further reading or examples?
I think that would be a very tall order. It could easily mistaken a number of other things as a cobweb because they are so small and they move. I just dust mine off every couple of weeks because I have the same problem, cobwebs triggering motion recording or interfering with a clear image of my target.
The only way you are likely to do this is to have a real-time image processor on an external computer with a lot of resources to process it. Even today, many of those systems still have issues identifying a dog vs a human vs a car - let alone a tiny web or spider on it.
I’ve seen a few mentions of integrating with LLMs so possibly something to replicate this manual test with my buddy ChatGPT-800
Possibly my prompt lead him to his conclusion but would be interesting to see how many false positives…
In this attempt below he seems to have got it correct so thats 2 for 2 so far 
