How to perform feature matching using GPU on Windows machine

Hello, I am having memory problem on my Windows machine while using feature matching function for a sequence of image data (resolution: 1920x1080). I am using ORB feature detector for feature detection and extraction. The machine I am using has 16 GM memory with Intel Core i7 @ 3.20 GH processor. I also have a GPU Radeon RX590 Series but I don't get how to use it for feature matching on Windows. The example given is only supported on Linux OS.

Please guide me how can I solve memory problem?

Thank you in advance.

[Walter Roberson]

Radeon is not an Nvidia gpu. Only Nvidia Cuda capable gpu can be used.

[jhz]

I have other machines as well with Nvidia GPUs but with lesser memory (8 GB). I would like to know how can I use Nvidia GPU on Windows machine for feature matching.

[Walter Roberson]

Sorry it is not obvious to me why that example is restricted to Linux support. I do see that TensorRT support is currently restricted to Linux but that requires compute capacity 6 and the example only requires 3.2 so it cannot be using TensorRT.

I seem to recall seeing that nvcc for Windows is not as stable as on Linux.

I do not have a licence for the gpu Coder toolbox. I also do not have a Windows development machine set up (Microsoft says volunteer work is not Personal Use and wants me to pay $7000 per year for full professional licenses.)