With the rapid development of artificial intelligence technology, AI painting shows revolutionary potential in the field of art creation. In this paper, we focus on the convolutional neural network-driven AI painting generation model (FSMNet), realize image classification and style migration based on VGG-16, and construct an AI painting generation algorithm adapted to art education. Experiments show that compared with StyTr2, which has the best performance in the baseline model, the SSIM and PSNR of FSMNet are improved by 18.6% and 6.82%, respectively, and the migration time is reduced by 0.017s to reach 0.51, 12.06, and 0.079s, respectively. After the teaching experiments, the experimental group scores in the three dimensions of drawing fundamentals, color perception, and creative thinking respectively reach 25.43±1.14, 27.09±1.28, and 35.18±2.15, which are all better than the control group, and the total score of the test is higher than that of the control group by 13.68, and the standard deviation is smaller, and the overall performance is more stable.
