Research on modern art creation trend and its visual expression based on deep learning technology

We propose the AI painting model based on the generative adversarial network and analyze the color, pattern, and visual hierarchy of the AI painting’s visual expressiveness. Starting from the painting sketch, the corresponding generator and discriminator are integrated to construct the AI painting model based on a generative adversarial network, and the training strategy, optimization objective, and loss function calculation formula of the model are given. The visual analysis of the AI painting based on deep learning technology is combined with research analysis. In terms of simplifying sketches and generating pictures, the model presented in this paper performs more prominently than the control models (CNN and SVM), which is reflected in the loss value and F1 value, respectively. In addition, there are significant differences between AI paintings and traditional paintings in the use of color (P = 0.035, T = 1.478), pattern cultural allegory (P = 0.011, T = 0.347), and visual hierarchy (P = 0.008, T = 2.993), and AI paintings are more visually expressive than traditional paintings, and at the same time, the regression equation of the AI paintings’ visual expressiveness is 0.125+0.179*Color Use+0.188*Pattern Cultural Allegory+0.234 Visual Hierarchy. This study provides a clearer perception of the modern art creation trend and its visual expressiveness, which is of enormous significance in boosting the development of the art field.