At least one processor determines a person state of a crowd according to a people state control designation as designation information on a person state of people and an individual person state control designation as designation information on a state of an individual person in the people. The at least one processor generates a crowd state image as an image in which a person image corresponding to the person state determined is synthesized with previously-prepared image at a predetermined size, specifies a training label for the crowd state image, and outputs a pair of crowd state image and training label.