The prevalence of camera networks in public places has increased substantially over the last ten years. This is in part due to improved vision algorithms and the development of simple, less expensive cameras. With this increased prevalence, a number of privacy concerns have been raised. Particular concern has focused on the level of detailed information that image data provides and the real time operation of most camera networks. In response to these concerns, policies and best practices have been suggested. In this work we look at current deployment policies and suggest technical solutions that enhance privacy awareness. We propose the use of system design measures, such as validation codes and online notices. We also look at different levels of data abstraction that can be performed using computer vision techniques, and characterize the information these methods provide about the scene, allowing practitioners to determine which data must be used for to support the network mission as well as the information that can be withheld. We provide experimental results on these measures and suggest open areas of research.




Download Full History