CSE 457 Final Project
Vihar Desu | vihar.desu@wustl.edu | 443845
Nick Murray | nicholasmurray@wustl.edu | 441567

27K

StreetStyle: Exploring world-wide clothing styles from thousands of photos

Each day billions of photographs are uploaded to photo-sharing services and social media platforms. These images are packed with information about how people live around the world.

In this visualization, we explore an annotated set of ~27,000 Instagram photos across 31 cities from Cornell University researchers (http://streetstyle.cs.cornell.edu/).

Our dataset is composed purely of annotations about clothing color, categories, the image and the upload coordinates.

We present a visual analysis of clothing and fashion across these images of people around the world spanning several years, from June 2013 to December 2015. These sample trends could be possible avenues of future exploration for the larger dataset of 10 million images analyzed by the Cornell researchers, or any other large fashion dataset.



Top 5 Trends: Bangkok

This chart displays the top five combinations of a color and clothing category (sweater, t-shirt, shirt, dress, tank top, outwear, suit and uncategorized) that are more popular in this city than the rest of the world. Each of the five groups comprises at least 1% of the overall items in a chosen city, and the multiplier underneath the group name displays how much more likely a style combination will appear in that city compared to the other 35 cities in the dataset. Mouseover for details.



Timeline Distribution of Items by Category and Color: Bangkok

This chart displays the timeline of all the items uploaded to that particular city. The same circles in the trends above are replicated below.

More than 1 color items are represented by a black circle with a red outline. Uncategorized colors are white with a silver outline.