search

TheImagingSource / Linux-tiscamera-Programming-Samples

Programming samples in Python and C++ for the tiscamera GStreamer modules.
71 stars 35 forks source link

High CPU load with Raspberry Pi Compute Module 4 / GStreamer #45

Open crl-manoj opened 2 months ago

crl-manoj commented 2 months ago

Hi, we are using tiscamera library with Raspberry Pi Compute Module 4. The gstreamer image capturing only 8 frames using python is talking lot of cpu usage for all four cores in the compute module 4. What might be the reason for this? It is using about around 60% CPU on all cores on average.

Camera Type: Model: DFK 33GX264 Serial: 46120683 Type: aravis

Distro: Distributor ID: Debian Description: Debian GNU/Linux 12 (bookworm) Release: 12 Codename: bookworm

tiscamera library: tiscamera: 1.1.1.4142 tiscamera-tcamproperty: Provided by tiscamera tcamdutils: 1.0.0.560 tcampimipisrc is not installed. tcamtegrasrc is not installed. tcamdutils-cuda is not installed. theimagingsource-drivers is not installed. ic_barcode is not installed.

TIS-Stefan commented 2 months ago

Hello I think, the PI4 is somewhat to weak for the image processing needed. I am aware, that sounds like I would like to make it easy for me, but unfortunately the PI4 is not really that powerful You also installed all necessary software, e.g tcamdutils, which is faster than bayer2rgb. You may try the standard pipeline gst-launch-1.0 tcambin ! videoconvert ! xvimagesink and see, how this works.

Stefan

crl-manoj commented 2 months ago

@TIS-Stefan I checked it out didn't see much difference. The issue happens when we are taking image frames snapshot. Do you have any recommendations for hardware for imaging intensive application using this camera?

TIS-Stefan commented 2 months ago

Hello

You may try a PI5 instead. But the CM5 is not available yet as I know. I also did not find time to run the GigE and USB camera on a PI5.

Stefan