I used the MTG for Unity to create a trained dataset with two models (a simple box and a more complex sculpture). It seems to work, both models show up in the console as "new search results" when I look at them with the camera. That means they got recognized, right?
But only one of them appears in "Trackable...found".
What is the difference between a recognized model ("new search result...") and a tracked model? When Vuforia detects a model, doesn't it automatically know where it is in the room?