Pattern Customization Using Neural Networks

Gustavo PÉREZ-CORONA, José Anibal ARIAS-AGUILAR, Moisés Emmanuel RAMÍREZ-GUZMÁN

Article
2026 / Volume 9 / Pages 1036-1052
Received 17 December 2025; Accepted 24 February 2026; Published 27 April 2026
https://doi.org/10.31881/TLR.2026.1036

Abstract
In the fashion industry, patterns play a fundamental role as they constitute the structural foundation of any garment. Although technological advances have facilitated the digital pattern-making process, this work still largely depends on the experience and knowledge of skilled designer-pattern makers. Training an expert pattern maker requires significant investment of time and resources from companies. In this context, we present an innovative approach for personalized pattern construction, supported by 15 years of experience on pattern making and machine learning programming. Our elaboration process for basic women’s patterns (front and back bodice) is based on industrial and custom patternmaking knowledge, using 13 structural measurements of the female body as input to a neural network model who generates a personalized pattern as output. The potential benefits of our approach are multiple. It can not only help nonexpert designers create personalized patterns but also improve time efficiency for experienced designers. Furthermore, our model facilitates the transition to industrial-scale applications, enabling companies to reduce their dependence on expert pattern makers and promote efficiency in the pattern-making process.

Keywords
pattern customization, neural networks, digital patternmaking