With the rapid evolution of artificial intelligence (AI) technology and the widespread adoption of multimedia applications, Art and design, as disciplines closely linked to contemporary trends, urgently require teaching models that overcome the deficiencies of traditional teaching models in terms of practicality and innovation. This paper is dedicated to developing a multimedia intelligent teaching method that integrates AI. By constructing a Multi-Scale Adaptive Style Fusion Network (MASF-Net), based on the StyleGAN2 framework, this method integrates a multi-scale feature encoder, a Content-Style Adaptive Weighting Module (CSAW), and a personalized interpolation strategy tailored to different user skill levels. This achieves efficient and precise conversion from sketches to artistic renderings. The test validation shows that the method significantly improves the generation quality, and the peak signal-to-noise ratio (PSNR) of the method reaches 24.8 dB, the structural similarity index (SSIM) reaches 0.85, and the FID value decreases to 40.2. Furthermore, the method has a high user satisfaction survey score of 4.3 out of 5. During the 8-week teaching practice, the creativity realization and technical expression ability of the students in the test group reaches 82.5 points and 81.3 points respectively, which are significantly improved compared with the control group. This study not only confirms the advantages of the proposed method in enhancing teaching interactivity and adaptability but also provides a practical intelligent solution for art and design education. It points out the direction for future development in lightweight design, small sample learning, and integration with augmented reality technology, and has important practical value in promoting educational innovation.
Zheng et al. (Fri,) studied this question.