How to compile Opencv with CUDA

Hi everyone,

I flashed my AGX Orin Dev Kit with SDK Manager and Jetpack 5.0 (CUDA and other frameworks included)
Completed this process 100% success and just made upgrades after.

However, CUDA doesn’t work with my Deep Learning and Computer Vision applications and not compiled with OpenCV. So it runs so slow with low fps. First of all, is that normal and expected situtation ?
If not, what am i supposed to do to compile CUDA with OpenCV or make my Dev Kit ready for deep learning applications.

Before AGX Orin, I used to use Jetson Nano for deep learning projects and satisfied with result. I had prepeared it with this kind of instruction " How to configure your NVIDIA Jetson Nano for Computer Vision and Deep Learning - PyImageSearch "

Will I do similar processes like creating Virtual Enviroment, Cmake compiling with compiler flags, using docker etc. ?

Other user has shared a script to manually build OpenCV. Please give it a try:
GitHub - mdegans/nano_build_opencv: Build OpenCV on Nvidia Jetson Nano

Would you share the cmake flags?

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.