Welcome to the Face Detection Data Set and Benchmark (FDDB), a data set of face regions designed for studying the problem of unconstrained face detection. This data set contains the annotations for 5171 faces in a set of 2845 images taken from the Faces in the Wild data set.