Representations for object and action recognition