The project aims to develop a model for identifying and distinguishing textile fibers from high-resolution images. For this purpose, an extensive database will be created, comprising high-resolution images of the four main types of fiber used in textiles – cotton, flax, wool, and silk – both in pure form and in various blends, based on samples from modern textiles. The images will be acquired using a microscope and systematically organized for the training of a machine-learning model.

The model will be trained to accurately identify the fiber type and, where possible, estimate the relative proportions of the fibers in a blended fabric. The final application will allow images to be acquired directly from textiles using either a microscope or a camera fitted with a macro lens, and the images will then be automatically analyzed by the model.

At the same time, a corresponding database will be created using textiles from cultural heritage objects, and an attempt will be made to validate and/or improve the original model for historical textiles. Differences in manufacturing techniques, deterioration, or ageing may create difficulties; however, the model will be adapted where necessary so that it can also be applied to such textiles, helping to protect them while providing valuable information about their fibers and condition.