I want to implement a custom CUDA kernel that has direct access to the image data. I am using a GpuMat, which has several built-in functions I want to use within my kernel such as rowRange(). See : https://docs.opencv.org/3.4/d0/d60/classcv_1_1cuda_1_1GpuMat.html.
Is there a way to do this? Or are all these functions only available from the host side, and I cannot access them inside my CUDA kernel in device?
Thank you.