Bibliographic citations
Quispe, H., Ching, S. (2024). Identificación de fibra textil natural mediante visión computacional y técnicas avanzadas de aprendizaje profundo [Trabajo de investigación, Universidad Peruana de Ciencias Aplicadas (UPC)]. http://hdl.handle.net/10757/673411
Quispe, H., Ching, S. Identificación de fibra textil natural mediante visión computacional y técnicas avanzadas de aprendizaje profundo [Trabajo de investigación]. PE: Universidad Peruana de Ciencias Aplicadas (UPC); 2024. http://hdl.handle.net/10757/673411
@mastersthesis{renati/416470,
title = "Identificación de fibra textil natural mediante visión computacional y técnicas avanzadas de aprendizaje profundo",
author = "Ching Donayre, Sui San",
publisher = "Universidad Peruana de Ciencias Aplicadas (UPC)",
year = "2024"
}
The raising of camelids in the high Andean regions of Peru is an essential agricultural activity. The alpaca represents the main source of income for these largely impoverished communities. Currently, the trade of artisanal textile products supports about 90,000 alpaca herders and 36,000 textile artisans. However, the presence of unfair competition, which is strengthened by the difficulties in guaranteeing the authenticity of textile products, generates distrust among consumers, negatively affecting these economic sectors. In addition, the identification of textile fibers, a slow and costly process that guarantees the authenticity of the product, requires highly qualified professionals. This situation makes it difficult for honest textile producers to guarantee standards of value and quality for their products. This research develops a predictive model to facilitate the identification of fibers in textile products using microscopic images. This model is a component of a device called Fibroscope, which will allow the capture of microscopic images and make real-time predictions without the need for highly qualified personnel. The research begins with the design of an image capture protocol, followed by the construction of an image set based on this protocol. Textile samples are provided by specialized institutions in the textile sector. Subsequently, the best predictive model is selected among the existing CNN architectures. This advancement will allow entrepreneurs and textile artisans to improve consumer confidence, democratize the sale of textile products, and increase the income of one of the productive sectors with lower incomes.
This item is licensed under a Creative Commons License