GPU support

[Gegam]

GPU support

[pvnieo]

Hi,

• The actual implementation doesn't run on the gpu, so even if you have a RTX3080, it won't be used.
• Bigger images are skipped because the optimization will take too much time, but you could not skip them and wait for the optimization to converge!
• I don't see why you need multiple gpus for this problem
• I you want to use gpu, you can convert this implementation to your favorite deep learning framework (e.g pytorch, tensorflow), and for that, you need to re-implement all the predefined functions available in OpenCV, like sobel filter and Mertens merge.

Hope this answer your question!