FaceBlock: Semantic Context-Aware Privacy for Mobile Devices R.Yus, P. Pappachan, P.K.Das, E. Mena, A. Joshi and T. Finin U. Zaragoza UMBC FaceBlock automatically obscures faces in pictures based on dynamic, context-aware policies Glass users Everyone else Using Google Glass for photography could make you a social pariah Being photographed without consent can be a worry Users take their picture and specify a policy regarding pictures taken by others, e.g., obscure me in pictures taken by strangers while not at work or after 5:00pm FaceBlock automatically generates an eigenface id, discovers nearby Glass devices running FaceBlock and exchanges information with them User’s device maintains context model and informs Glass device when to block FaceBlock makes respecting personal privacy as easy as using Google Glass itself Automatically opt out of pictures based on context dependent policy rules FaceBlock on Glass responds to other devices, receiving eigenface ids and block status messages Faces in pictures are identified and compared to active eigenface ids When a face is matched, its block status is checked and the face obfuscated as needed A simple protocol supports device discovery and data exchange Contexts are modeled in RDF supported by OWL ontologies Supported by NSF #0910838 & #1228673, CICYT TIN2010-21387-C02-02 and DGA FSE http://face-block.me