80 Million Tiny Images
Here’s a beautiful web project: 80 Million Tiny Images, a visualization of all the nouns in the English language arranged by semantic meaning, by Antonio Torralba, Rob Fergus and William T. Freeman at MIT.
It’s not just pretty, either: the project aims to use this massive dataset to train a computer to recognize objects within an image and to understand the scenes depicted in photographs. What’s fascinating to me is the way the composite image comes out so familiar – colours of the earth, I suppose. This is despite the fact that each of the tiles in the “poster” represents a composite image of one of the 53,463 nouns (retrieved from WordNet) representing an average of about 140 images (a total of about 7,527,697).
Is it art, is it linguistics, or is it science? The answer of course, in the traditions of the best projects, is all of the above.