Exploring a Makeup Support System for Transgender Passing based on Automatic Gender Recognition

Title of the Paper

Exploring a Transgender Makeup Support System Based on Automatic Gender Recognition (AGR)

Paper Information

• Subject Area: Human-Computer Interaction and Applications of Diversity and Inclusion Technologies
• Keywords: Automatic Gender Recognition (AGR), Transgender, Makeup Support, Human-Computer Interaction (HCI), Gender Affirmation, Inclusive Design, Virtual Makeup, Technology Ethics, Japanese Culture, Data Privacy

Research Background and Issues

• Problems and Challenges:

  • Automatic Gender Recognition (AGR) systems perform poorly when applied to transgender populations and may lead to gender discrimination issues.
  • Transgender individuals in non-Western cultures, such as Japan, face unique challenges, such as a lack of makeup learning resources.
  • Current virtual makeup systems primarily target cisgender women and fail to address the needs of transgender individuals.

• Significance:

  • Unconscious biases in AGR technology can cause psychological stress and even discrimination risks for transgender individuals.
  • Supporting gender identity expression for transgender individuals through technology in culturally diverse contexts has significant practical implications.

• Research Motivation and Related Work:

  • The authors aim to explore whether AGR systems can transition from being tools of negative control to tools that positively support transgender practices.
  • Previous studies have largely focused on Western countries like the United States, lacking data and case studies from non-Western cultural contexts.
  • Current virtual makeup and recommendation systems lack designs tailored for transgender individuals.

Solution

• Methods and Innovations:

  1. Technical Innovation: Develop a virtual makeup system, "Flying Colors," for transgender individuals using AGR technology, allowing users to upload photos and explore how makeup affects their "gender passing ability."
  2. Design Insights: Specifically designed for the Japanese transgender community, emphasizing the influence of cultural context on makeup behavior and gender identity expression.
  3. Key Features:
     • Makeup Feedback: Provides instant feedback through the AGR system, assessing the impact of specific makeup styles on gender passing ability.
     • Makeup Recommendations: Automatically generates five optimized makeup styles based on feedback scores.

• Implementation Steps:

  1. Conduct a three-year digital ethnographic study of the Japanese transgender community to identify issues and needs.
  2. Develop a prototype virtual makeup system, integrating commercial virtual makeup components with AGR-based scoring.
  3. Perform preliminary testing and evaluation of the system, collecting feedback through interviews and user experiments.

Research Outcomes

• Specific Findings:

  1. Participant Survey:
     • Investigated the makeup habits of 15 transgender participants, their views on passing, and their acceptance of AGR systems.
     • Most participants expressed neutral to positive attitudes toward the system's scoring feature.
  2. Experimental Evidence:
     • AGR systems can support users in exploring makeup to enhance gender passing ability or achieve broader gender identity expression.
     • Participants noted the system's potential value in improving makeup skills and exploration, especially in safe, private environments.

• Advantages:

  • Offers a low-risk way to explore makeup without external judgment, avoiding potential societal gender discrimination.
  • The system is designed to quantify gender passing ability, enabling transgender individuals to set personal goals.
  • The "blameability" of AGR makes it easier to accept when results do not align with expectations.

• Experimental and Evaluation Results:

  • The system's scoring feature received an average rating of 4.8/7, considered relatively "reliable."
  • The makeup recommendation feature scored 5.3/7, indicating areas for future improvement.

• Limitations and Future Directions:

  1. System Limitations:
     • Current AGR systems face accuracy and bias issues, such as inadequate support for non-binary genders or specific ethnicities.
     • The realism and customization options of virtual makeup templates need further improvement.
  2. Potential Ethical Issues:
     • AGR systems could be misused to externally evaluate transgender individuals' gender passing ability.
     • There may be negative impacts on users' mental health and makeup decision-making.
  3. Future Improvements:
     • Add non-binary and more personalized makeup options, considering the needs of transgender men.
     • Design specialized datasets for different cultural or background contexts to improve the applicability and accuracy of AGR systems.
     • Further study the long-term psychological effects of system use and establish stricter privacy protection and informed consent mechanisms.

Conclusion

This paper explores the potential of AGR systems to support makeup and gender identity expression for the Japanese transgender community and designs and tests a prototype virtual makeup tool. The study demonstrates that, under strict privacy protection and user choice, the tool can effectively and satisfactorily support transgender makeup learning. However, careful balancing of ethical considerations and technical efficacy is required in the design and application of such technologies.