Second-hand shops are booming and giving many items of clothing a second life. But sorting old clothing is time-consuming, which is why many clothes that are still usable end up in the trash. Researchers are therefore working on optimizing this process. A newly developed process based on artificial intelligence could help to automatically sort old clothes and textile waste. In the future, the AI-supported scanner will also record what material the clothes are made of and whether they are contaminated with harmful substances.
The fashion industry with its trend towards “fast fashion” causes an estimated eight percent of global CO2 emissions. Clothing therefore does not have a particularly good carbon footprint. As the population becomes more climate conscious, the demand for second-hand clothing is increasing. Flea markets, online sales platforms and even large fashion chains offer previously worn clothing. However, used textiles that are no longer wearable in Europe are either recycled into wiping cloths, painter's fleece or car seat fillings or exported to African countries. A lot of things just end up in the trash. The European Union wants to change that and ban the export of used clothing.
In order to better utilize domestic textiles, Europe will have to sort even more in the future. However, this is time-consuming and personnel-intensive. “So far, old clothes have been sorted by hand. Every piece of clothing has to be handled two to three times until the type of product and its quality are determined,” explains Karsten Pufahl from the Technical University of Berlin.
AI based scanner records clothes and helps with sorting
In order to make the second-hand business much more attractive, a research team led by Pufahl, consisting of scientists and employees from waste disposal companies, has now developed a textile scanner that records clothing items in detail and thus helps with sorting. The scanner first photographs items of clothing on a black conveyor belt. The technology then uses artificial intelligence to analyze within seconds what type of product it is and whether the item of clothing is suitable for the local second-hand market or should be recycled.
The scanner not only records whether it is trousers, a blouse or a coat, but also what color the item of clothing is, what condition it is in, what brand it comes from and whether it corresponds to current fashion. In order to assess this reliably, the researchers regularly train the AI with data about current trends. This effective sorting should result in more textiles ending up on the second-hand market and more items of clothing finding a second life. Not only is this more sustainable, the team reports that most of the money can be made with old clothing.
New process detects material and pollutants
In the future, the used clothing scanner will be able to recognize even more features and thus enable even better recycling of the textiles. Pufahl and his colleagues are currently researching how the AI system can better assess the quality of the clothes and how it can identify what material they are made of. With this knowledge, old textiles could be used to produce more high-quality yarn in the future, which could then be used to create new textiles - in keeping with the circular economy. This so-called fiber-to-fiber recycling has not yet been used much because the near-infrared spectroscopy currently used can only be used to say with certainty for a few tissues what material they are made of, the scientists explain. These include white bed sheets or jeans that are made of just one fiber. However, most clothing consists of fiber mixtures that cannot easily be made into new yarn without being separated.
In order to accurately detect these materials, the researchers are now using a more advanced form of so-called Raman spectroscopy, which can precisely detect any type of textile. Theoretically, it can even be used to measure certain pollutants in textiles. However, scientists are currently having difficulties finding useful training material for the AI. “It is complicated to both obtain the pollutants and to find clothing contaminated with them,” says Pufahl.
Source: Technical University of Berlin