Smart camera networks provide the opportunity to detect, classify, and trace visual events by means of a wireless network of embedded computing devices equipped with camera sensors. Previous research in this area has largely focused on custom hardware solutions. In contrast, we propose to use mobile phones as a portable and lowcost platform to implement smart camera networks. Due to mass production, the price of mobile phones is constantly dropping while additional functionality is being added. In particular, the addition of cameras and ad hoc networking enable the use of mobile phones for smart camera networks. Also, software developed for mobile phones is highly portable due to standardized programming environments and APIs. In this paper, we present Facet, a self-organizing network of smart cameras and its software architecture, which addresses the specific challenges of the mobile phone platform. An initial prototype of Facet has been developed and is used to obtain first performance results in an office setting.