Clustering Embroidery Images: A Comprehensive Analysis of Methods to Track the Road of Embroidery Spread

The paper will reveal how contemporary technologies, like machine learning could help in determining how the embroideries found in traditional Bulgarian clothing are spread among ethnographic regions in the time. The Bulgarian embroidery also known as "shevitsi" is deeply rooted in the country’s cultural heritage. Different regions of Bulgaria have their own distinct embroidery styles, reflecting local traditions, influences, and histories. Tracing these patterns helps in documenting and preserving this regional diversity, which is crucial form maintain the cultural richness of the country. In the paper we have investigating different unsupervised machine learning techniques to cluster images in the dataset based on common features such as colors, motifs, patterns, etc. The paper presents a comprehensive workflow for analyzing Bulgarian embroidery traditions as found in folklore clothing. The analysis suggests that the presence of detail metadata with georeferencing and data annotation, would significantly enhances the ability to trace the spread and evaluation of embroidery motifs. This dataset would allow for a thorough temporal analysis, enabling us to observe how motifs and colors have evolved over time.