Eyes on the past

Version 2

Faces offer an instant connection to history, reminding us that the past is full of people. People like us, but different. People with their own lives and stories. People we might only know through a picture, a few documentary fragments, or a newspaper article.

On this site I’m exploring whether faces can provide a way to explore more than 120 million newspaper articles available on Trove.

To create the site I filtered and downloaded thousands of illustrated articles from 1880 to 1954 using the Trove API. I then used OpenCV to look for faces within those illustrations, and eyes within the faces. I only kept the faces in which I'd been able to detect an eye. The result was about 1800 images. A small and unevenly-sampled collection, but enough to build something with.

The interface was built using Flask, MongoDB, JQuery, Isotope, and Bootstrap. It’s hosted on Heroku and, for the first time, I'm storing all the images on Amazon S3.

About 100 randomly-selected faces are displayed at a time. Reloading the page should change the selection. Click and explore. They’re watching and waiting...

Please cite as: Tim Sherratt, Eyes on the Past, 2014, <https://dx.doi.org/10.6084/m9.figshare.4315697.v1.>