DeepStack and CPU processor question

[dsabot]

I am currently running 20 4-8mp cameras in our warehouse.
We have DeepStack turned on for all cameras with:
Min confidence: 50%
real-time image: 25
analyze one each 250ms

The PC specs are:
Intel i5-9500 @3GHz
16 G of ram

Before we implemented DeepStack, we were at about 25% load at peaks.
Now we are at 100% and 3G ram.

I am sure Deepstack is causing this performance drop, probably due to all the motion in our warehouse (people constantly moving).

Any suggestions on tweaking some settings?
We are trying to record 24/7 (stream 2 / 640x480) and switching to 4-8mp upon DeepStack image confirmation

 

[wittaj]

You have to either tighten up the motion trigger, not run it on as many cameras, or add a GPU to the computer to offload DeepStack to.

You can turn off the Use mainstream for DeepStack because Deepstack downrezes anyway. Substream should be fine. I found setting Deepstack to Low instead of Medium or High is just as accurate and less resource intensive. YMMV.

And make sure you have done EVERY optimization in the wiki:

Optimizing Blue Iris's CPU Usage

Blue Iris uses a lot of CPU time. In most Blue Iris installations, your CPU is the limiting factor which determines how many cameras you can have. As such, it is extremely important to optimize your configuration to ensure you can get the most out of your system while consuming the least...

ipcamtalk.com

 

[iwanttosee]

Add a Nvidia GPU. The more cuda cores the better. I choose a P620 for low wattage and low proflie size, it has 512 cuda cores.

 

[dsabot]

You have to either tighten up the motion trigger, not run it on as many cameras, or add a GPU to the computer to offload DeepStack to.

You can turn off the Use mainstream for DeepStack because Deepstack downrezes anyway. Substream should be fine. I found setting Deepstack to Low instead of Medium or High is just as accurate and less resource intensive. YMMV.

And make sure you have done EVERY optimization in the wiki:

Optimizing Blue Iris's CPU Usage

Blue Iris uses a lot of CPU time. In most Blue Iris installations, your CPU is the limiting factor which determines how many cameras you can have. As such, it is extremely important to optimize your configuration to ensure you can get the most out of your system while consuming the least...

ipcamtalk.com

Thanks! Do you mind sharing what you have min confidence / + real-time images and analyze one each ms set to?

 

[wittaj]

Each field of view is different. DS doesn't go below I think 40%, so anything below that is not needed.

I would suggest 60%, 2 additional realtime images and 750ms apart and see if that gets you what you need. Obviously the more additional images and the closer the time, the more CPU intensive it is. And then go up or down from there based on how it is performing.

 

[dsabot]

Thank you @wittaj I made your suggested tweaks and it dropped our CPU load down to the mid 60s. I think part of the issue was the aggressiveness of the number of images and how quickly I asked DeepStack to review. Will work on tweaking some more to get this all locked down.

 

[wittaj]

Awesome! I would think in a warehouse setting you have more ideal situations light-wise that you don't have to be as aggressive as trying it at night.