It seems strange that they would include in their set of example images, a picture of the most famous mausoleum in the world, without it being tagged with mausoleum or tomb or anything like that.
And it is tagged 99% mosque, while it isn't one. (the building on the left of it, not in the image, is).
If I uploaded my own picture of the Taj Mahal and it told me it was a Mosque, I wouldn't be surprised, and I'd probably be reasonably impressed. The dome and minarets do rather give that impression, and I wouldn't really expect a computer to be able to tell the difference.
The reason I find it odd is that I would expect the first example on a demo to be carefully chosen to show off the system in the best light. It would be one that has perfect or near-perfect tagging. Maybe later on, I would show the shortcomings with a tricky image like this.