Writing performance

[DelgadoPanadero]

I have seen the code and it seems that the schedule_frame() method is a perpetual disk writting of images in the virtual node path, am I right?

I was wondering if there is any solution that let a python process to "send" processed (for example a deep learning detection algorithm) to a virtual device path which can be retrived by another python process on demand (for example an api, with reads the stream when a certain method is call) without having to write the frames in memory