DataFace: left_behind

Greg Dorsainville

DataFace: left_behind is an installation piece that captures the diversity of faces we see when we consume commercial media. Through the process of making, I will examine themes that include bias in algorithms and the nature of diversity.


The installation is designed for a general audience in a space with walls for large projection. It is a series of data visualizations projected on objects and the wall. By using computer vision face detection (and in the process understanding and improving on its default irregularities), I will extract faces from a corpus of commercial media from broadcast television.

This data I am calling a face diet and the installation will show who makes up this diet. The data is projected on objects placed on a tabletop using projection mapping. Animation of this data will be used to demonstrate other themes of the project.

The projection scale of these large datasets gives an opportunity to explore the nature of diversity in broadcast media.