We aim to revolutionise and automate textile sorting for a circular economy

The Project

  • Within the framework of CRTX, TU Berlin, circular.fashion and Freie Universität Berlin are researching a solution that aims toward automating the sorting of used garments and textile waste for high quality purposes in a circular economy.

    Our goal is to close the gap between the collection of used textiles and specific sorting for second hand and fiber-to-fiber recycling by means of AI-backed spectroscopy and image analysis, thus enabling a continuous material cycle. High quality recycling techniques such as fiber-to-fiber recycling require material specific feedstock with high purity to recover yarn of the same quality. So far, available sorting solutions have not delivered the required precision to allow recyclers to work with post-consumer textile waste.

    CRTX brings together a team of experts in the field of optics and data-science and sustainability to develop a new data-driven multisensor sorting solution to solve this challenge.

    Reusable textiles, on the other hand, will be processed with methods of computer vision to support sorting personshuman sorters achieving a more fine grained and objective classification. Using a deep learning with state-of-the-art training methods the team develops solutions that are optimized for conveyor belt garment identification. To make second hand products more attractive and curated according to the target groups, their needs and wishes curate and carefully select products in a more targeted way put the for second-hand markets in a better position, trends from the first-hand market are analyzed and constantly matched with the sorting algorithm. This allows to target more fashion-aware consumer groups, increasing the overall usage of second-hand garments.

    A market analysis of the first-hand identifies trends and matches these with the detailed descriptors as prevents fiber-to-fiber recycling, which would allow yarn of the same quality to be recovered. This is one of the main reasons for the low recycling rate (less thanaround one percent) of textiles.

    The spectroscopic methods, which are being developed under the direction of Dr. Karsten Pufahl from the Department of Nonlinear Optics, should enable an exact determination of the material composition and pollutant load.

    An AI-based evaluation is planned, which will overcome the previous hurdles in sorting technology and enable a closing of the material cycle. At the same time, image analysis methods are to be used to achieve more precise sorting, also for reusable garments.

    circular.fashion will coordinate the development and contribute essential expertise and partner networks from the industry.

Intelligent textile sorting

How it works

project-thumb
precise and fast

Textile spectroscopy


Material detection

Sorting for fibre recycling: Using high-speed raman spectroscopy we want to deliver material analysis for highly precise recycling feedstock
project-thumb
data-driven approach

AI methods


Product classification

Sorting for second hand use: With AI backed computer vision we classify product types to find the optimal target group and client for reusable garment to give more garments a second life
project-thumb
Phasing out hazardous substances

Chemicals database


Clean feedstock

Our high resolution spectroscopy is trained to find textiles containing hazardous substances to take these garments out of the loop
project-thumb
helping sorters with

Sorting Recommendations


Trend based sorting

We watch the first hand fashion markets to identify trends and help the second hand retailers to react immediately to customer demands.