AMNESIA (automated methods for numerical, easy, secure and irreversible anonymization) is an in-house developed software based on a multistage algorithm for detecting and anonymizing faces and license plate numbers found on photos (pixilated). Automatically-generated suggestions can also be edited.
AMNESIA was created as part of a master’s thesis by the accident research/interior concepts team under the leadership of Christian. The motivation for the software was two-fold, the first of which was related to ensuring data privacy and observing artistic copyright laws. A second point was time savings. For each accident that is investigated, our colleagues make between 300 and 600 photos that must be manually anonymized after the fact. This equates to up to three hours of work for each analyzed accident.
Other optimizations to the software are in the pipeline, such as expanding it to video/moving picture material for wider use in experimental fleets.