In this paper, a natural approach of the describing for the kids which it is brought into use, by making the image engraving technique of Deep Learning into thought. The previous work opens the technique for a course semantic mix plan (CSF) to mine the specialist features to encode the image content through thought instrument. A CSF benefits by three sorts of visual thought semantics which including the thing level, picture level, and spatial thought features, in three-stage course way. In the viewpoint on the current structure, the model takes the moved picture given by the customer as the information and starts recording the image what a PC find in human standard language with point by point explicit presents in the image appropriately exhibiting the significance of picture captioning. This strategy can be applied to the field of youngster’s tutoring to make them pattern of adapting also entrancing. Fundamentally, we facilitate thought instrument with three kinds of features to orchestrate the setting data about the photos from differentviewpoints.