Question:

[mcorner]

@chuckcho I am interesting in PTZ detection for use in a research project.

Two things:

1. Did you ever implement zoom detection?
2. Is this based on any particular published algorithm, or is it just your own heuristics?

This seems to be one paper on the subject:
https://www.cse.iitb.ac.in/~sharat/icvgip.org/ncvpripg2008/papers/9.pdf

And it does something kind of similar: "The number of optical flow vectors that
correspond to the dominant orientation should contribute to
more than n% of the total number of vectors, if the motion
was induced due to pan change. "

But that isn't quite the same.

Thanks for the code!

[chuckcho]

@mcorner 1. No. haven't implemented zoom detection. 2. This is out of my own little experiments on a handful of samples. So, this may not generalize well for different data samples.