Tutorials on detecting black screen from live video frames

I am looking for any tutorials/guidance on detecting black screens from live video frames. Any help on this is appreciated.


1 Like

I could imaging that taking the sum of all pixels of the current frame and using a simple threshold might work already assuming that “black screens” have (mostly) zero pixels while “non-black screens” contain some color information.


Many thanks @ptrblck for your response. lets say if we want to detect anomaly screen on web UI testing, would you suggest auto-encoder approach or classification approach for this problem statement?

I don’t know what “anomaly screen on web UI” would mean and how it would “look” so cannot recommend a specific algorithm to detect it.
E.g. if the “anomaly” looks like a white noise screen, you should be able to use the entropy as a feature and applying a simple threshold might already work quite well. On the other hand, if the “anomaly” would change e.g. the text of a web UI, the use case could be much more complicated.