During the initial sorting of the collected materials, one question keeps coming up: Is this even a textile?
In SorTex’s collection tests, we see that a significant portion of the incoming material is actually not textile. These may be shoes, bags, belts, or other items that have been turned in along with the textiles. It is important to identify and remove these products early in the process. If they travel too far through the sorting system, they can affect material quality and create challenges in subsequent steps.
Preliminary findings suggest that this initial separation is one of the most critical steps in the entire sorting process.
AI as a Tool for Initial Sorting
To better understand and support this process, SorTex is working on the development and testing of AI-based object recognition. The setup is relatively simple: A camera is positioned above a conveyor belt and analyzes the material flow in real time.
Although the test images may often appear neat and orderly, the models are trained to handle mixed and unpredictable material flows that reflect the reality of collected textiles.
At this time, the model is working, among other things, to identify:
- Non-textile products such as shoes, bags, and belts
- Small or complex textiles
- Worn, damaged, or contaminated textiles
Better Decisions Throughout the Value Chain
The purpose of classification is not merely to sort materials. It is also about creating a better basis for decision-making later in the process.
When materials are correctly identified at an early stage, it becomes easier to assess the most appropriate next steps, whether they involve recycling, redesign, alternative use, or material reuse. Conversely, value can be quickly lost if materials are placed in the wrong streams.
It may sound like a minor detail, but our experience shows that these early decisions, in particular, have a significant impact on the quality and efficiency of the entire circular value chain.
