ZoneMinder Forums

I have searched for some time for (dummy-proof) documentation on how to correctly set up remote object detection and machine learning on a separate server with no real luck. Hoping someone might have some insight or at least a general guide for how to do this. I did find this thread but haven't ...

Possible? Yes. You didn't say whether you are doing passthrough or encode. Encoding takes a ton of ram too (default settings will buffer 20 frames in the encoder). Passthrough can buffer some stuff but it is encoded data and should just be ref counted so shouldn't amount to much. --I'm using ...

iconnor wrote: ↑Mon Mar 22, 2021 3:12 pm So you guys who are seeing this ram use, I need debug level 4 logs for zmc.

Forgive my ignorance - where/how can i get this information to give you?


Dan

I'm noticing that the memory issue is only happening when the cameras are set to record of any kind, and then drops back off when the event(s) are complete and the process starts over again (~10 mins) . I also noticed that there are 4 zmc processes per camera enabled. Could this also be attributing ...

this was resolved by installing scikit_build with pip3 (not pip)


Dan

So, I have ZMES, MLAPI and ZM all working independently, but have not found a solid reference to integrate MLAPI into the ZM installation so that the objects identified are in the monitor.