A semantic reconstruction and AI-controlled generation method for the cultural genes of Qing Dynasty embroidery patterns: a case study of official rank badges
Year: 2025
Type: article
Abstract: As an important carrier of traditional Chinese culture, Qing Dynasty embroidered rank badges (Guangbu) hold rich cultural genes and historical information within their patterns. Currently, in the research and innovative design of traditional patterns, there are still multiple challenges, including digital expression, cultural gene extraction, and modern application. This paper, for the first time, constructs an 'embroidery semantic network' for Qing Dynasty embroidered rank badge patterns, achieving structured expression of pattern cultural elements.It also innovatively introduces 'dynamic shape grammar' to achieve precise control over pattern forms. Combined with a three-stage fine-tuning architecture of LoRA-Diffusion-SG, it significantly enhances the realism and cultural alignment of generated patterns. The research integrates multi-source data such as images, atlases, and literature, establishing a comprehensive dataset comprising image, knowledge, and craftsmanship layers. Experiments show that the proposed method performs excellently in terms of cultural compliance (CLIP similarity 0.78–0.82), symbol recognition accuracy (Top-1 82%–85%), and craft feasibility prediction (F1-score 0.87–0.89). his study provides a theoretical basis and technical support for the digital preservation and redesign of embroidery patterns, promoting the deep application of artificial intelligence in cultural heritage inheritance.
Source: npj Heritage Science
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Funder: Guangzhou University
Award: G602854899
OpenAlex ID: W4417101608