VorlonCD/bi-aidetection

disadvantages of not using queue?

Opened this issue · 5 comments

I had added more cameras for AI detection and started getting max limit queue errors.. I increased it to 150 and still get max limit error from time to time and it seems to blocks all detection of future images.
I have disabled queue on the camera and deepstack settings and I don't seem to have any issue since then.
What are the disadvantages having disabled these 2 queues?

You cant disable the image queue, just change its size. You can disable the server queue and action queue, maybe thats what you are thinking?

You have a few solutions:

  • Make the queue much larger. But then it can take much longer to get an alert.
  • Reduce the number of images being created by blueiris - Make the motion detection less sensitive, or on the record tab reduce the "each m:ss.ss" setting. A few more settings may affect how long images are created for but I forget all of them.
  • Add more machines on the network with Deepstack servers to process the images faster so the queue doesnt get so big. You could try running more instances of deepstack on the same machine using different ports but I've found it gets less stable and of course slower. The best way is to use separate devices.

yeah I meant the server queue and action queue.
I don't have any error since turning them off (probably mainly due to server queue). What are the disadvantages of having disabled the server queue? will there be any side effects?

hover over checkbox for description

https://i.imgur.com/miERP76.png

I guess there's no side effects and it seems to be it's even better having it unchecked?

If you uncheck it, and have more than one AI server, I would be the fastest one at the top