A Practical System for Privacy-Preserving Video Surveillance

Video surveillance on a massive scale can be a vital tool for law enforcement agencies. To mitigate the serious privacy concerns of wide-scale video surveillance, researchers have designed secure and privacy-preserving protocols that obliviously match live feeds against a suspects' database. However, existing approaches provide stringent privacy guarantees and, as a result, they do not scale well for ubiquitous deployment. To this end, we introduce a system that relaxes the underlying privacy requirements by giving away some information when a face is compared against the law enforcement's database. Specifically, our protocol reveals a random permutation of obfuscated similarity scores, where each obfuscated score discloses minimal information about the actual similarity score. We show that, despite the relaxed security definitions, our system protects the privacy of the underlying faces, while offering significant improvements in terms of performance. In particular, our protocol necessitates a single round of communication between the camera and the server and, for a database of 100 suspects, the online computation time at the camera and the server is 155 ms and 34 ms, respectively, while the online communication cost is only 12 KB.