Skip to main content
RISE logo

Artificial intelligence gives less wastage in the fashion industry

With the help of more efficient and precise analysis, players in the fashion industry can take smarter and faster decisions that reduce excess production and the impact on our climate. Decisions can be made about what investments they should make, when they should make the investments, the size of the purchases they need to make, and how much needs to be produced. AI can bring great benefits to rise above competitors in the industry, but above all, AI can create the conditions to reduce wastage in the fashion industry.

The fashion industry is currently undergoing major challenges and many of the prime Swedish fashion players are experiencing difficulties adjusting to meet the new demands from society and consumers. While digitalisation offers opportunities to reach new customer groups and increase sales, organisations need to change their processes and structures. Several of the fashion industry players have had to introduce extensive cost saving programmes that have resulted in shop closures and reduced staff numbers.

The fashion companies' analysts have an increasingly important role in the strategic work of the organisations. The analysts are looking for trends and flows in the fashion world before they reach our market. The further in advance they can discern a new trend and identify the size of the trend the better. Analysts' insights and data then form the basis of the design and purchasing organisations' decisions on next seasons' investments.

AI facilitates for fashion industry analysts

Olof Mogren and John Martinsson, both researchers at RISE, are currently conducting a large project to facilitate the fashion industry's analysis work using artificial intelligence and deep learning.

“Trend analysis today consists largely of following blogs and social media, but we hope that the results of our project will provide techniques that facilitate and streamline the work of analysts,” says Olof Mogren.

“With the help of semantic segmentation, we can identify which garments a person in an image that is spread on the internet is wearing, how they wear the garments and how ordinary the garments are. AI is used to facilitate the analysis of large flows of, for example, fashion images, where the processing of such a large volume of images from a resource standpoint cannot be handled manually,” explains John Martinsson.

Technically, semantic segmentation means that an image is divided into the different objects contained in the image. When it comes to fashion images, it is of interest to divide them into parts that depict different types of garments, jewellery, bags, backgrounds and the like. The image is processed in a neural network that classifies each pixel. The semantic segmentation marks the pixels as one of several objects and in this way the system can identify and distinguish different details of the image.

The project ends in June 2019. With the new technology, the researchers hope that AI can help fashion companies to better forecast their purchases and production and thus reduce overproduction in the industry and the impact on the climate.


The research group consists of Olof Mogren, John Martinsson and Abubakrelsedik Karali from RISE and Oskar Juhlin from Stockholm University. The Swedish Fashion Council are also participating in the project as well as major players in the Swedish fashion industry.