FaceID@home: cycle-sharing for facial recognition

Over the last years we experienced a revolution in the creation and dissemination of media content such as videos and photos. However, this information is not cataloged and there is no automatic system that indexes movies according to its content. To solve this problem we propose FaceID@home. This system will automatically detect and recognize faces using the Eigenfaces algorithm. To tackle the high demand of computing power from video and image processing, we will execute this system in large scale through a public distributed computing infrastructure based on BOINC. We will also make use of the volunteers’ graphics processors to improve the speed of facial recognition algorithms. With this work we expect to create a free, efficient and scalable system that automatically identifies and indexes people in multimedia content.