Collection and Analysis of a Representative Dataset of Human Skin Colors

Team: Bolanle Dahunsi, Justin Geeslin, Heidi Woelfle, Lucy Dunne

Program: Wearable Technology Lab

What is the range of numerical values within which human skin colors fall and does the RGB or LAB space represent it more compactly?

The fashion industry struggles with diversity of representation, and white female models tend to be overrepresented. The ongoing dissatisfaction with this trend has led to efforts to include more people of color in fashion. As representation improves, designers are beginning to consider a broader range of potential customers. However, the breadth of human skin tones is not well-characterized in a way that can be used by designers. This study therefore sought to clearly define the spectrum within which human skin colors fall using the RGB and LAB spaces, and understand which system provides a more clearly defined range for human skin colors. Images from cabinet members from 32 countries across the globe were collected and average skin color values for each image were extracted using photoshop in both the LAB and RGB spaces. The numerical ranges identified here helps build on ongoing work to improve personalization of apparel recommendations based on individual physical features.

Funders: National Science Foundation