A 3D Dynamic Database for Unconstrained Face Recognition

In this paper, we present a new 3D dynamic face dataset dedicated to the development and test of algorithms which target face recognition under unconstrained conditions, from 3D videos. Several challenges which can occur in real world like scenarios are considered, such as continuous and freely-pose variation, expressive and talking faces, changes of the distance to the 3D camera, occlusions and multiple persons in the scene. In this database, a full 3D static model is collected for each subject, together with eight 3D video sequences. Each video lasts about 20 seconds, including challenging variations under continuous and freely pose variation. Single-view structured-light 3D scanners are used in the acquisition process. This dataset contains 58 subjects. To provide baseline recognition performance on this database, a 4D-to-4D subspace learning face recognition approach is introduced and experimented. To the best of our knowledge, this database is the first 3D dynamic database designed for the purpose of face recognition considering freely-moving 3D faces. As such, it provides to the research community a new benchmark that can stimulate investigation of face recognition algorithms under new and challenging conditions.