I have a python script that opens the camera and processes each frame, i have tried a few different gstreamer pipelines but none of them helped me increase the fps, I have been trying to get high fps at higher resolutions but when i set the resolution to 1080p the fps gets very low, I have noticed that the cpu is only being utilized one core at a time and the gpu is hardly used, how could i make it so the script utilizes the gpu or utilizes more cpu cores at the same time? I have tried the python script on my laptop and the fps is higher than it is on the Orin NX.
Jetpack 5.1.2
Orin NX 16GB
Python 3.8
Opencv 4.5
Camera= Raspberry pi HQ
hello yigit.muhittin.57,
please check the sensor capability, $ v4l2-ctl -d /dev/video0 --list-formats-ext
you may also try below gst pipeline with disable preview and shows frame-rate only,
$ gst-launch-1.0 nvarguscamerasrc sensor-id=0 ! 'video/x-raw(memory:NVMM),width=1920, height=1080, framerate=30/1, format=NV12' ! nvvidconv ! 'video/x-raw(memory:NVMM),format=I420' ! fpsdisplaysink text-overlay=0 video-sink=fakesink sync=0 -v
hello, yes the sensor capability at 4k is 30fps and at 1080p is 60fps, when i try the gst pipeline i dont have any problems but as i said before i want to process each frame with my python script, i am trying to do object detection, with the gst pipeline you gave it doesnt use the cpu too much, how could i achieve that in my python script aswell?
hello yigit.muhittin.57,
so, that’s not an issue with camera stack.
may I know which sample you’re using?
you may try with deepstream samples, which are located at… /opt/nvidia/deepstream/deepstream/samples/
besides, please see-also Performance — DeepStream documentation 6.4 documentation for reference, thanks
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.