This is just a quick memo to myself on how to visually detect security cameras. Written down some commands, thoughts and the weights I generated along the way.
With the YOLO library building these types of solutions to visually detect objects seems to be relatively straight forward. Here is an example of it working, having used a rather small sample set and a medium sized model:
Would be nice to integrate this into some fancy glasses with a camera somewhere in the future. For now it is good enough to use on recorded video material.
DiabloHorn 