For those of you that don't know, this article is (probably?) indebted to the 2011 SO question How do I find Waldo with Mathematica?[1]. It was a pretty big deal at the time and I find it odd that the Wolfram article doesn't mention it anywhere (unless I missed it; I just skimmed the blog post).
“...who sat for three hours in a Barnes & Noble bookstore with a measuring tape painstakingly constructing this fabulous dataset.”
Wow, couldn’t actually buy the books?!
Seconded.
I've always wanted to train a little neural net image classifier to find Waldo's face. It seems like that would be a more useful general approach to finding Waldo in any image. However it would break down on the last Waldo challenge in one of the books where you have to find Waldo in a world of Waldo lookalikes based only on the fact the real Waldo is missing a shoe.
Some trivia: In the rest of the English speaking world (including the UK where the series of books originates from) the series is known as Where's Wally: