"Security Tool – Privid – Guarantees Privacy in Surveillance Footage"

Researchers from the Massachusetts Institute of Technology's (MIT) Computer Science and Artificial Intelligence Laboratory (CSAIL) collaborated with other institutions to develop a system called Privid that ensures the privacy of video footage from surveillance cameras. Privid is a privacy-preserving video analytics system aimed at supporting aggregation queries that process large amounts of video data. The system allows analysts to submit video data queries, and adds noise (extra data) to the result to ensure that an individual cannot be identified. The team's system builds upon the formal definition of privacy, known as differential privacy, which enables access to aggregate statistics on private data without exposing Personally Identifiable Information (PII). Analysts would typically have to access the entire video to do what they need to do with it, but the Privid system ensures the video is not a free-for-all. Honest analysts can access the information they need, but with the Privid system, that access is restricted to the point where malicious analysts cannot do too much with it. In order to enable this, Privid breaks the video into small pieces and runs processing code over each chunk instead of running the code over the whole video in one shot. Rather than getting the results back from each piece, the segments are aggregated, and additional noise is added. For example, the code could output the number of people observed in each video chunk, and the aggregation might be the sum. Privid allows analysts to use their own deep neural networks, which are common for modern video analytics, thus giving analysts the flexibility to ask questions that Privid designers did not expect. Across various videos and queries, Privid was accurate within 79 to 99 percent of a non-private system. Officials could use Privid to collect secure public health data or allow transportation departments to monitor the density or flow of pedestrians without revealing personal information about people. This article continues to discuss the development, capabilities, and uses cases of the Privid privacy-preserving video analytics system.

SciTechDaily reports "Security Tool – Privid – Guarantees Privacy in Surveillance Footage"