Online Archive of University of Virginia Scholarship
A Proposed Best-Case/Worst-Case “In The Wild” Analysis of Few-Shot Human Image Transfer Approaches; Gender as a Sociotechnical Construct301 views
Author
Rothenberger, Jay, School of Engineering and Applied Science, University of Virginia
Advisors
Ku, Tsai-Hsuan, EN-Engineering and Society, University of Virginia
Campbell, Brad, EN-Comp Science Dept, University of Virginia
Abstract
As an engineer, I am responsible for not only my actions, but for the actions of the things that I create. Societies are built upon and mediated by technologies created by engineers. Nowhere is this mediation stricter than online. In the digital space we can only interact with the help of technology. In an effort to ensure this mediation is fair for everyone, I have proposed an experimental evaluation of recent and promising approaches to human appearance transfer. Approaches like these are of concern for communities that engage in gender exploration, because they are an essential part of virtual clothes try-on implementations, and facial recognition, among related computer vision applications. These computer vision applications serve as an integral part a powerful tool for individuals to explore alternative modes of physical or digital expression, whether that be visualizing oneself differently in a physical space or transforming a digital model or image in a digital space according to one’s movements.
Towards better understanding the community I hope to serve with my technical research, I conducted interviews and participant observations of individuals who have performed gender experiments and online communities for gender exploration respectively. My research yielded a discussion that includes some recommendations for engineers that seek to include or serve gender explorers with their digital platforms. My research also showed that in our hybrid digital-physical world gender is plainly constituted as a sequence of acts rather than biology, which has additional implications for facial recognition software that seeks to classify individuals based on gender.
Degree
BS (Bachelor of Science)
Keywords
Gender, Computer Vision, Butler, Latour
Notes
School of Engineering and Applied Science
Bachelor of Science in Computer Science
Technical Advisor: Brad Campbell
STS Advisor: Tsai-Hsuan Ku
Technical Team Members: Jay Rothenberger
Rothenberger, Jay. A Proposed Best-Case/Worst-Case “In The Wild” Analysis of Few-Shot Human Image Transfer Approaches; Gender as a Sociotechnical Construct. University of Virginia, School of Engineering and Applied Science, BS (Bachelor of Science), 2021-05-12, https://doi.org/10.18130/g21c-az28.
