Design and Development of Modified Swin Transformer Based Digital Cultural and Creative Designs for Enhancing the Style Transfer Mechanism

Digital cultural and overwhelming creative designs has attracted more significant intensity of momentum in imparting digital practices such as art, music and cultural heritages. These practices need more intelligent algorithms for the enhancing the transferring the mechanism. The recent advancements in Machine learning and Deep Learning algorithms have shown remarkable potential in analysing and generating culturally rich visual contents. However, achieving the high accurate performance in existing models remains to be real challenge to an efficient style transfer and exact art generation. This research article explores the advantages of Swin transformer and proposed Enhanced Swin Transformer Model integrated with Deep Weighted Feed Forward Networks to support the cultural design tasks including style classification and segmentation process. The suggested Swin transformer model was trained and validated with Kaggle Digital Art and Cultural datasets which contains $\mathbf{1 0, 3 6 5}$ cultural images which is useful for enabling multi-model cultural representation. The experimental framework evaluates performance of method through metrics like PSNR, SSIM, DICE, accuracy, precision, recall and $\mathbf{F 1}$-score are calculated. The suggested model has achieved PSNR of 35.92,SSIM of 0.92, DICE of 0.93 and accuracy of 0.987 respectively. The study demonstrates that proposed mdoel not only facilitates cultural preservation but also fosters innovative creative practices, bridging the gap between heritage and modern design.