Research on optimization and style control of convolutional neural network in complex pattern generation model of paper-cutting art based on AIGC

Abstract

Background: A traditional intangible cultural heritage, paper-cutting art is also endangered by the diminishing numbers of practitioners as well as the transition toward digital art. With their recent ability to digitize and innovate this art form, artificial intelligence (AI) specifically the convolutional neural network (CNNs) have risen up as such a transformative tool. Through this systematic review, we examine the use of AI to optimize and control styles for creating complex paper cutting patterns, which simultaneously protect for cultural authenticity while cultivating modern creativity.
Objectives: In this paper, we review the optimization and style control mechanisms used in CNNs to increase the efficiency, accuracy, and cultural relevance of AI generated paper cutting patterns. Moreover, it also focuses on larger uses of AI in the field of cultural preservation, as well as the main challenges, and presents recommendations for future research.
Methods: Multiple databases, including Scopus, IEEE Xplore and SpringerLink, were searched for studies published between 2010 and 2024. Inclusion and exclusion criteria were established to select studies that used AI or similar methodologies to work on paper cutting, or other cultural art forms. The PRISMA guidelines were followed in data extraction and quality assessment. A detailed analysis was provided based on findings from 12 studies.
Results: Pattern generation fidelity and efficiency were significantly enhanced by optimization techniques including: perceptual loss functions, adaptive algorithms, and wavelet analysis. With style control mechanisms, the design principles of traditional symmetry and abstraction were preserved, while simultaneously allowing for innovation. Virtual reality (VR) among these immersive technologies made education and engagement interactive for wider audiences. But even though the datasets are limited, the computation demand is high, and there are ethical concerns around cultural representation, these challenges require collaboration.
Conclusion: Preserving and modernizing traditional paper cutting art, instigating global cultural exchange and innovation can all be achieved through the use of AI technologies. By bringing interdisciplinary collaboration and expanded datasets to bear on addressing challenges, we can do so in culturally sensitive and accessible ways. This review offers the role of AI in keeping the tradition and modernity of intangible cultural heritage.