ARCHIVES

Managing and retrieving embroidery patterns from large textile datasets is a challenging task in the textile and fashion industry. Traditional embroidery design searching methods are manual, time-consuming, and inefficient for handling large collections of patterns. To address these challenges, this paper presents the design and implementation of an AI-based Embroidery Design Matcher Using Deep Learning system for intelligent embroidery pattern retrieval using image-based, text-based, and hybrid search methods. The proposed system utilizes the CLIP model for generating semantic image and text embeddings and employs FAISS (Facebook AI Similarity Search) for fast and efficient similarity matching in high-dimensional vector spaces. Additionally, OpenCV-based color and texture feature extraction techniques are integrated to improve retrieval accuracy through hybrid similarity scoring. The system also incorporates a Stable Diffusion-based Generative AI module for creating new embroidery patterns from text prompts. The implementation provides a secure and user-friendly web application with efficient data management and scalable retrieval performance. Experimental results demonstrate that the proposed system improves embroidery pattern search accuracy, reduces retrieval time, and provides an effective solution for textile and fashion industry applications.