Data Driven Patternmaking: Automated Generation of Stretch-Knit Patterns for Different Fabric Stretch

Abstract

Patterns serve as the blueprint for garment production; therefore, accurate patternmaking is crucial for producing well-fitted garments with minimal trial and error. Patternmaking is time consuming and requires specialized expertise and proficiency. Current research in automated patternmaking has increasingly focused on enhancing pattern precision through computational methods. Recent developments have addressed data driven modeling and machine learning in automated pattern creation. Yet, the influence of fabric stretch on pattern adjustment has not been studied. This study examined how fabric stretch characteristics apply to garment fit and developed predictive models to create pattern shapes of womenswear upper torso stretch garments. The research objectives were to review existing data driven approaches for automated patternmaking techniques, create CAD patterns based on the sizes and fabric stretch values, develop a machine learning model to predict pattern shapes, and validate the predicted patterns through digital fit assessments. Adopting a machine learning approach, this study applied multiple linear regression to predict pattern alterations based on size and fabric stretch values. Furthermore, the same framework is improved to automate the pattern generation with fabric stretch values and size inputs provided. The findings contribute to the development of more accurate, automated patternmaking systems that account for fabric stretch, enhancing efficiency and fit consistency in apparel production.