Privacy Protection for Large Scale Naturalistic Driving Videos

A common pool of naturalistic driving data is necessary to develop and compare algorithms that infer driver behavior, in order to improve driving safety. The SHRP 2 naturalistic driving study (NDS) is currently collecting such data for the past two years which will result in approximately 4 petabytes of data, 1 million hours of video, 3000 subjects, 5 million trips, 33 million miles driven and 4 billion GPS points [1]. The Transportation Research Board, however, is working on details to give public access to this data due to personal identifiable information and protection of privacy. We propose and implement de-identification filters for protecting the privacy of drivers while preserving details to infer driver behavior, such as the gaze direction, and show promising results.