Question about expanding streams (scaling).

[epigraphe]

Search before asking

• I have searched the Supervision issues and found no similar feature requests.

Question

Now I am using your example, processed for launching on streamlit. Everything works as expected, the camera sees people and counts them.
Now I need to scale the system. Roughly speaking, on the home page of Streamlit I will have 6 cameras, each of which should detect people and display their current number.
Is this possible within one server? If I simply copy a working recognition block 6 times within one script, will everything work together?
I understand that this will take a large amount of resources, we are ready to purchase video cards and other power for this, but the question here will be, will this work in principle?

Additional

No response

[SkalskiP]

Hi, @epigraphe 👋🏻 Do you run inference locally or over the API?

[epigraphe]

Everything works on one machine and with one script. I don’t want to flood and I won’t copy it again))), I already posted an example of the script earlier, here.
#390 (comment)

Now it works on a CPU with an 8n model, we plan to transfer it to a GPU

[onuralpszr]

@epigraphe If you want to run for 6 camera what you can do is If you wanna do via "ultralytics" you
can do "multhread-tracking" example explain here or you can setup a inference server and call via that as well.