非语言差距:朝着更安全、更公平的在线约会的情感计算机视觉

📄 中文摘要

在线约会已成为浪漫关系的主要起点,但现有平台剥夺了人们依赖的非语言线索,如目光、面部表情、身体姿态和反应时机,这些线索用于传达舒适度、兴趣和同意,从而造成了沟通差距,尤其对女性的安全产生了不成比例的影响。这个差距既是计算机视觉领域的技术机遇,也是道德责任。计算机视觉社区已经开发出情感工具,如面部动作单元检测、目光估计、参与度建模和多模态情感识别,但在约会领域的研究上却大多被忽视。提出了一项以公平为首要目标的研究议程,旨在填补这一差距。

📄 English Summary

The Nonverbal Gap: Toward Affective Computer Vision for Safer and More Equitable Online Dating

Online dating has emerged as the primary means for initiating romantic relationships; however, existing platforms eliminate essential nonverbal cues such as gaze, facial expressions, body posture, and response timing that individuals use to communicate comfort, disinterest, and consent. This results in a communication gap that disproportionately affects women's safety. This gap presents both a technical opportunity and a moral responsibility for the computer vision community, which has developed affective tools like facial action unit detection, gaze estimation, engagement modeling, and multimodal affect recognition. Despite these advancements, the dating domain has largely been overlooked as a research context. A fairness-first research agenda is proposed to address this gap.

Powered by Cloudflare Workers + Payload CMS + Claude 3.5

数据源: OpenAI, Google AI, DeepMind, AWS ML Blog, HuggingFace 等