The Selfie dataset contains 46,836 selfies and annotates 36 attributes including gender, age, race, face shape, expression, hair color, decoration, and others. The dataset was originally used for the task of selfie scoring, but because of its wide dimensionality, it is very suitable for related research.