Facial recognition (FR) technology is being adopted in both private and public spheres for a wide range of reasons, from ensuring physical safety to providing personalized shopping experiences. It is not clear yet, though, how users perceive this emerging technology in terms of usefulness, risks, and comfort. We begin to address these questions in this paper. In particular, we conducted a vignette-based study with 314 participants on Amazon Mechanical Turk to investigate their perceptions of facial recognition in the physical world, based on thirty-five scenarios across eight different contexts of FR use. We found that users do not have a binary answer towards FR adoption. Rather, their perceptions are grounded in the specific contexts in which FR will be applied. The participants considered a broad range of factors, including control over facial data, the utility of FR, the trustworthiness of organizations using FR, and the location and surroundings of FR use to place the corresponding privacy risks in context. They weighed the privacy risks with the usability, security, and economic gain of FR use as they reported their perceptions. Participants also noted the reasons and rationals behind their perceptions of facial recognition, which let us conduct an in-depth analysis of their perceived benefits, concerns, and comfort with using this technology in various scenarios. Through this first systematic look into users’ perceptions of facial recognition in the physical world, we shed light on the tension between FR adoption and users’ concerns. Taken together, our findings have broad implications that advance the Privacy and Security community’s understanding of FR through the lens of users, where we presented guidelines for future research in these directions.

从确保人身安全到提供个性化购物体验的各种原因，私人和公共领域都采用了面部识别（FR）技术。但是，尚不清楚用户如何从实用性，风险和舒适性方面看待这一新兴技术。我们开始在本文中解决这些问题。尤其是，我们在Amazon Mechanical Turk上与314名参与者进行了基于小插图的研究，以基于在FR使用的八种不同情况下的三十五个场景，研究他们在物理世界中对面部识别的看法。我们发现用户对于采用FR并没有二进制答案。而是，他们的看法是基于将应用FR的特定上下文。参加者考虑了多种因素，包括对面部数据的控制，FR的实用性，使用FR的组织的信任度以及FR的位置和周围环境，以便将相应的隐私风险置于上下文中。他们在报告自己的看法时，将FR使用的可用性，安全性和经济收益权衡了隐私风险。与会者还指出了他们对面部识别的理解背后的原因和理由，这使我们能够对他们在各种情况下使用该技术的感知利益，关注点和舒适度进行深入分析。通过对系统用户在现实世界中对面部识别的感知的首次系统研究，我们了解了帧中继采用与用户关注之间的紧张关系。在一起