``I am not the primary focus" - Understanding the Perspectives of Bystanders in Photos Shared Online

When taking photos in a crowd, unintended individuals, such as bystanders, are often captured alongside the main subject(s). In an effort to protect bystanders' privacy, existing methods have been developed to automatically detect bystanders. However, inconsistent definitions of who qualifies as a bystander limit their effectiveness. To better understand bystanders' perceptions, we conducted an online survey with 486 participants, analyzing their responses to 864 image-based scenarios and their comfort with sharing these images online. Our results revealed no significant correlation between comfort with public photo sharing and bystander status. We identified limitations in current bystander detection methodologies, as they often fail to recognize bystanders who are not clearly in the background, hence missing individuals with privacy concerns. Moreover, comfort with public sharing varied significantly depending on the image context. Our findings highlight the importance of considering the context of captured images to address privacy concerns in image sharing.