AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )
http://pjreddie.com/darknet/
Other
21.65k stars 7.96k forks source link

CUDA not found jetson xavier agx #8495

Open ashsifat opened 2 years ago

ashsifat commented 2 years ago

I am building in my jetosn xavier agx, the makefile works, but i need to use the cmakelists.txt to be able to integrate with other projects. I get this error, I have specified the cuda 10.2 library path export PATH=/usr/local/cuda-10.2/bin${PATH:+:${PATH}} export LD_LIBRARY_PATH=/usr/local/cuda-10.2/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

and disabled vcpkg integration, still no effect. Any idea how to solve? TiA

CMake Warning at CMakeLists.txt:105 (message): vcpkg not found, toolchain not defined, system not win32 so build might fail

-- CUDA_PATH: -- CUDACXX: CMake Error at CMakeLists.txt:154 (message): CUDA not found, please build explicitly with -DENABLE_CUDA=OFF if you do not want CUDA.

-- Configuring incomplete, errors occurred!

image

stephanecharette commented 2 years ago

Define:

to be able to integrate with other projects

Because the usual way to build on Jetson with the Makefile works fine, and the resulting libdarknet.so and header file is extremely easy to integrate with other projects.

For example, this is how I do it and recommend to other people that they do it: https://github.com/stephanecharette/DarkHelp#how-to-build-darkhelp-linux

ashsifat commented 2 years ago

Thanks for the idea Stephan, I am not an expert in cmake and make, I would have to use this library with ROS packages which uses Cmakelists.txt, so I thought building with the Cmakelists.txt would be a straightforward way to do it. But I am looking into the way you suggested and seeing if I can make it integrated with projects that use Cmakelists.txt with cpp codes in .cc files formats.