Semantically-sensitive images are segmented and objects are classified according to a combination of color, texture, size, shape