Democratizing Facial Recognition with Google Glass

Abstract

Lightweight and camera-equipped wearable devices such as Android-backed Google Glass— with their potential for wide-spread and mobile data capture—have piqued the imagination of technologists and privacy advocates alike. This paper describes an experimental system which confirms the feasibility of such devices for surveillance through live data collection and facial recognition. Furthermore, even though effective surveillance tasks are computationally demanding, this work illustrates that performance of such systems is scalable through careful architecting of communication between static servers and mobile collection devices. When the bulk of the complexity can be offloaded to the server, and with the availability of highly-available communication channels between collector and processor, we have the foundation upon which future surveillance systems might be constructed. Such systems awaken nightmares for those advocating privacy of the modern citizen, while inspiring innovators to push forward the bounds of what can be accomplished with today’s technology. The present project enables advocates from both ends of the spectrum to debate privacy policy as it can be seen through the lens of systems that are possible today.