In this paper, an autonomous aesthetics-driven photographing instructor system is proposed, which gives instructions to help camera users to take good images. There are two kinds of instructions: image composition and personality feature enhancement. As for composition, a salient region is used to match the aesthetical template. To keep the personality of the autonomous photographing instructor system, the correlation between user types and image features is extracted through data mining on social networks, which is called “personality prediction.” The proposed system is run in realtime and workable on all mobile devices with cameras.