An Exploration of Culture through Unstructured Data

Mean Images is a Data Visualization project, implemented as a website. Inspired by Jason Salavon's work, our group realized that mean images could be an interesting tool to visualize images as non-tabular data. Therefore, we chose a variety of topics, gathered images with a web crawler, then compiled and narrated our findings as an interactive slideshow.

What's a 'mean image'? Simply speaking, it's an 'average photo' from a group of photos. To go into more detail, for a group of images of the same size, we can go from pixel to pixel, calculate the average color, then compile the averaged pixels to a new image. Though this 'averaging' might not make sense if it's applied to a random set of images, interesting patterns and trends emerge if the images are well-chosen.

Github repo