tri optique de textiles / optical sorting of textiles

The project aimed to propose a preliminary architecture of an optical sorting system for the development and integration into a value chain of an automated clothing sorting system. The value chain related to the sorting and recycling of clothing involves technological, economic, and social challenges. This project aimed to develop the technological aspect in order to facilitate the establishment of the ecosystem needed to build this value chain, from the collection to the valorization of textiles.

Conclusions of the project:

The technique of near-infrared spectroscopy is very popular as demonstrated by numerous publications. It is robust, easy to implement, and produces high-performance machine learning algorithm classification results. The spectroscopic and algorithmic measurement tools developed and used have been very effective for the optical sorting of textiles. Classification rates of 100% on samples of pure wool, silk, cotton, polyester, and mixed fibers were obtained in this preliminary study. Thanks to the performance of spectral and algorithmic analysis, doubts were raised about the composition of several samples.