<![CDATA[As an Indonesian cultural heritage recognized by UNESCO, batik features various motifs laden with philosophical values, yet public knowledge about batik patterns and their significance remains limited. This study presents a mobile-based batik classification system integrating MobileNetV2 architecture with a Retrieval-Augmented Generation (RAG) chatbot to provide interactive learning experiences, enabling users to identify batik patterns through image recognition while obtaining detailed information via conversational AI.This study adopts MobileNetV2 considering its efficiency on mobile devices. This model achieves an optimal balance between accuracy and computational performance. Model was trained on a balanced dataset of 5,000 images covering five pattern classes (Parang, Truntum, Kawung, Mega mendung, and Merak), achieving training accuracy of 98.97% and testing accuracy of 96.8%. The RAG-based chatbot, orchestrated using LangChain and Qdrant, enhances user interaction by retrieving relevant information from a curated knowledge base, ensuring contextual factual responses about batik's history, philosophy, and cultural significance. React Native was adopted as the development framework to ensure cross-platform operability. This implementation contributes to cultural heritage preservation by making batik knowledge more accessible through modern technology, combining computer vision and natural language processing in a unified platform.]]>
Copyrights © 2025
