Machine Learning for Classification of Bamboo Woven Design

Applications of bamboo woven design in the craft industry around the world are predominantly informed by highly trained artisanal skills. The research is conducted by a combination of primary and secondary data collections. Primary data collection includes collecting typical patterns from a design library and an interview with an expert bamboo artisan. Secondary data collection serves as the point of departure of the dataset, the 14 identified patterns. The primary objective of this study is to facilitate an accessible entry point for novice designers to engage with bamboo woven design, thereby preserving the essence of traditional bamboo weaving craftsmanship. Evaluation of the model through key metrics indicates the attainment of high accuracy and precision values. However, there exists an opportunity for future enhancement focused on improving the moderate 'loss' metric, elevating the recall value, and striking a delicate balance between precision and recall values. These aspects serve as the focal points for forthcoming studies, aiming to further refine the recognition system and enhance its utility for aspiring designers in the realm of bamboo weaving patterns.