Industry camera and opencv program on TX2

I am compiling opencv and industry camera sdk on Jetson TX2 to perform some real-time image processing, the way like ordinary compiling process on PC.

I would like to ask if the program can enjoy the GPU computation just to run it on TX2 or the program needs to be specially compiled with CUDA? Is there anyway to run the program with GPU acceleration? Thanks.

On Jetson platforms, we support hardware acceleration in gstreamer and jetson_multimedia_api. For using OpenCV CUDA filters, you may refer to the samples:
Nano not using GPU with gstreamer/python. Slow FPS, dropped frames - #8 by DaneLLL
LibArgus EGLStream to nvivafilter - #7 by DaneLLL

Thank you for reply.
I am using TX2 with the latest Jetpack installed. To use CUDA version opencv in C++, do I need to build new opencv with CUDA or it is already preinstalled in the Jetpack?

For some reasons, the opencv version shipped with JetPack does not have CUDA support.
If you want to use opencv CUDA modules with GpuMat, you would have to build by yourself. See these scripts.

1 Like

Honey_Patouceul is correct. Please try the script to build it manually.