Using OpenCV CUDA functions on THCTensor

vision
1

Hi,

I am trying to create my own layer in PyTorch based on some GPU based image processing. Fortunately, the function I am trying to implement is already available in GPU module of OpenCV.

What I do not understand is if and how I can use it directly to create the C code base for my own layer.

Can anyone please help with that ?

2

In most cases it is easier to reimplement the function in python using torch functions than working directly on THCTensors. Could you tell us what functionality you want to implement in your layer?