Object Recognition Performance & Best Practices?

Hi All,


I've started to prototype using the object recognition and while it is working, the performance results haven't been that great.


E.g. I scanned an object (Child's toy, a baby walker, which is mostly rigid with lots of contrast) and it seems to work very well in the original area I scanned it, but when I move it to other areas of the house, it has a great deal of trouble recognizing the object.


The lighting is slightly different, but not in a way that I would expect to throw off recognition.


Has anyone had any luck with object recognition producing consistent results? Anyone have any best practices out there?