I want to train a neural network with the info from images. I thought it made sense to use ORB.
As default ORB has got 500 features, which is a bit much to feed the network. How do you do when you want to train a network with images? I couldn't find any code where it was done before, so I need some help to figure out what I should feed the network...