Opencv 3.4.3 compilation error

Accelerated Computing CUDA CUDA Programming and Performance
#1

I am using OpenCV 3.4.3 which I compiled from sources using Visual Studio 2017. When I link to the compiled static lib to an application I get this error:

Error LNK2019 unresolved external symbol __cudaRegisterLinkedBinary_54_tmpxft_0000773c_00000000_17_gpu_mat_compute_70_cpp1_ii_71482d89 referenced in function “void __cdecl __sti____cudaRegisterAll(void)” (?__sti____cudaRegisterAll@@YAXXZ)
I am not able to figure out where it is coming from.
I looked online for similar errors and someone suggested to not use OpenCL, that did not work.

#2

Build your static lib with relocatable device code, and build your application with relocatable device code.

An example of the project setup to do this is here:

https://stackoverflow.com/a/54259206/1695960

#3

I am getting this error when compiling Opencv SDK samples. These samples have project files that are not CUDA projects in Visual Studio. Is there a way to handle this error in a regular windows VS project?

#4

For what its worth, check out this OpenCV + CUDA Module video walkthrough and tutorial. Hope this helps!

https://cudaeducation.com/opencvcuda/

-Cuda Education
cudaeducation.com