In the pursuit of visualizing 8000 images from the 2016 runways, I came across an interesting discovery about the uniformity of runway models. We all know the stereotype of models seemingly all coming from the same mold. By sorting these images by skin color and aligning their eyes, the extremity of the lack of diversity in the high fashion world is made evident. Nearly 8000 images blend into one face. The last 10-15 seconds of this four minute video are the only portion of the video where we start to see women of color.
I started by scraping 23,000 images from Vogue, and filtering out certain images (photoshoots not on runways, men, duplicates, etc.) In the end I was left with approximately 8000 images. I then tracked the face in order to derive average skin color, and sorted them by brightness. I created a program to record location of the left eye when clicked on. This allowed me to align all of the models by their eyes, and create a video of them ordered by brightness of skin.
Below is a Tsne grid created through ofxTsne showing 300 of these images grouped by like-ness. This Tsne was not as telling as I would have liked it to be due to the fact that in runway shows, sometimes the background is just as much a part of the show as the clothing being displayed.