Lead Author: Emily Long

Featured heritage: The Oddy Tests performed at the Metropolitan Museum of Art in New York, and some tests from English Heritage.

What we learned: It is possible to train a computer to classify Oddy test coupons using images. However, you need very good images, obtained with a consistent protocol. You also need a very good training set, which means that we first need to agree in which coupons have passed or failed the test! This paper is written in a very accessible language (as far as machine learning papers go), so it can be used as a way to become familiar with the field.

Where can I read it? The paper is available for free here