Anonymizing Audiovisual Data

The Audiovisual data present with Red Hen is not shareable. The visual data clearly shows the speakers involved, and a person can be recognized with the recorded audio. These datasets cannot be shared easily to protect the privacy of people. Thus, Red Hen needs an anonymization engine for its audiovisual data. Using audio processing techniques, classical computer vision algorithms and recent deep learning algorithms, we propose an anonymization engine. The engine will have a simple web-app interface, where users can choose the type of anonymization desired, or simply choose to randomly anonymize.