How to setup ML for testing computer monitor output?

Hi all,
new to ML, but have some experience in the past with ai software;

Goal now is to develop an app that can identify which screen is shown for test purposes.

usb video camera aimed at computer monitor from another pc. Need this to discern what screens are being shown.

Planned to setup fixture, use camera to walk thru pages and capture images that are fed to the ML framework (somehow) and get the result about which of say possible screens is active?

Am struggling to get a good video interface to this usb camera, purchased a Microsoft 3400 usb cam, have VS2019 community, idea was to just get to where ML can tell me which screen is active? Maybe set the camera to 5 frames/sec? and it can hopefully tell which screen it sees with say 85% assurance or?

So far the only ML add in was for F# which I know zero about, and already have other code in the c# win Form project. I know Azure can do it, but the cost seems like it would be prohibitive

Need info asap, thanks for any help or ideas!