Hi,
In a .cu device code, I pass cv::gpu::gpumat to kernel a through cv::gpu::PtrStepSz.
To do that I need to include <opencv opencv2.hpp=""> in the device code. In Windows I can compile such code in a cuda project including opencv.
Since nvcc is not able to compile opencv, how can I compile such code in linux?
My operating system is ubuntu14.04 with Cuda 6.5 and Opencv 2.4.12. Thanks!