Femme-Makeup Application as a Tool to Undermine Facial-Recognition Technologies
Goals: Explore how recent experimental makeup trends that feature traditional feminine aesthetics could be used to undermine facial recognition systems.
This is a new project and topic area for me, but it aligns with my interest in femme-tech. Due to increased concerns over the advanced state of facial recognition algorithms to target diverse bodies, from average citizens (link) to female sex workers (link), I’ve been considering tools to undermine these systems. For me, the leap to creative makeup as an avenue for undermining facial recognition systems was inspired by the CV Dazzle project (see Figure x).
Ultimately, at the completion of this work, I would like to have generated a more robust set of guides for applying creative makeup that prioritizes feminine makeup expression that is incredibly popular on social media (see Figure y). The aesthetic styles I’d like to explore are more traditionally feminine and often very artistic, as opposed to the CV Dazzle lookbook, which centers on an alternate aesthetic style. I think it’s interesting to work backward from looks that makeup artists are organically exploring to create facial recognition system guides, versus creating makeup looks for the purpose. Arguably this strategy could potentially be more viable for adoption.
I’ve found a handful of highly talented makeup artists that want to support this work. Conversely, one area where I could use more support is on the facial recognition technology side. I’ve found databases and algorithms online that I’ve run a sample data set through. The test resulted in 10% of the artistic makeup images not being recognized as faces by the software. I’d like to partner with someone with more experience working with facial recognition technologies to make sure I’m using a robust enough algorithm to prove the merit of my hypothesis.
I find the concept of traditionally feminine rituals and practices, such as the application of makeup, being used as defense measures against tools that have been used to disproportionately target women as an exciting research space.