Automatically identifies the visual center subject outline in the input image, erases the subject as the foreground and returns the segmented foreground subject image (4 channels), suitable for keying scenes of people, animals, food, objects, etc.