I'm currently using the human_pose_demo algorithm on my RPi 4, with the Intel Neural Stick 2, to detect person. Even with the stick, the program is consuming 80% (85% if I display the video) resources of the CPU. Is it normal?
Thanks in advance,
I am using a Pi 4 ( 4GB RAM and running Pi OS) with OpenVINO distribution 2021.1 and Logitech C925e USB camera. My CPU usage is approximate 48%.
Can you try running on a video file and compare to see if the Pi Camera is the cause of high CPU usage.
Sorry for the late reply.
I tried with a video (and also an USB camera), and I got the same result.
I got that CPU usage from htop, but if I do Ctrl+C on the video display, a CPU usage graph appears telling me 40 % (about same as you). So from where do you get your percentage?
I tried with a freshly install Raspberry Pi and OpenVino, to see if there was no interference.
I used top to get the CPU usage information.
When you said you get the same result, do you mean 80+% or the 40+% result?
Another check on the usage and it does almost add up to ~80% as I had chromium and some other application open on the Pi.
However, the demo itself should only consume around 40+ % CPU.