AI redefines the fashion industry
While it is beginning to be deployed in clothing design, textiles and sales, the integration of AI in garment making is still limited. But that could change...
Artificial intelligence (AI) could revolutionize fashion. "This technology is now mature. We're no longer just in the research phase, but in the industrialization phase. We can go confidently, there are no more technological locks" assures Robby Dubus, head of development at Textil'IA, a start-up created in 2020 by the textile profession to take advantage of new technologies like AI. The idea? To develop applications that respond to issues such as the demand for digitalization, the fight against counterfeiting, the enhancement of creativity...
Fast upstream
Photo credit : ©IMKI">As highlighted in the article "Texprocess 2024 goes AI", which appeared in the last issue of M&T2, this technology was the star attraction at the latest edition of the leading international trade fair for the processing of textile and flexible materials."AI has found its place and is developing very rapidly upstream: namely, the design of a product with all the parameters this entails, as well as its preparation. The same applies downstream, right up to the point of sale. But for the "hard" part, manufacturing and especially sewing, it's more complicated", stresses the magazine. Fashion is thus increasingly interested in the potential of AI. But at successive stages of the chain -creation, production, distribution- pioneers are more or less numerous to have already tested its possibilities.
Design at the cutting edge
Frédéric Rose, President of IMKI. Crédit photos : ©IMKI">Fashion is at the cutting edge of design, as explained by Frédéric Rose, founder in 2020 of IMKI, a start-up specializing in image-generating artificial intelligence solutions dedicated to all professions in the ICC (Cultural and Creative Industries). "When we worked on these industries, the sector that reacted most quickly was fashion. This one indeed needs AI to specialize in Technical fashion design".
To fine-tune its solutions, IMKI relies on both AI and professions experts. "AI doesn't repeat things. We take AIs that have learned to draw, we improve their training on professions and as a result, the accuracy of their answers. We are the only ones to have such an approach", asserts the executive.
For fashion, the start-up uses data-scientists and engineers and its four fashion designers / technical fashion designers (women's, men's, sportswear or "360 degrees" ready-to-wear) to train the AI to design garments. "We also involve professions specialists, as for a denim capsule presented at VivaTech 2024, improved by an external expert."
Time savings assured
Crédit photos : ©IMKI">IMKI takes "made-to-measure" a step further for its customers, such as The Kooples, with whom it has developed a capsule for the winter 2024-2025 collection. "We teach AIs to understand an artistic corpus, the codes to take into account to align designs with a brand. This could be the logo, a motif, a typical clasp for a handbag. In this way, we improve AI's knowledge by integrating the brand's DNA". This also avoids a risk to which IMKI's customers are very attentive: "making sure that the data they use to train their AI is not third-party data, but their property". To give them control over the AI, IMKI is thus training its customers in "prompting", i.e. how to give it instructions.
The question that arises: can AI replace humans for design? For some tasks, but not all, says Frédéric Rose. "AI is unbeatable in terms of speed of execution of a precise drawing, for example, but although hyper-powerful, this tool is nonetheless not endowed with willpower or a "spirit of synthesis" of the zeitgeist or trends. It needs creative input. If the designer doesn't know which artistic sphere to take it into, AI won't come out with anything good".
"AI speeds up the design process, improves decision support and brings the product closer to the final trend and customer demand. Today, without it, it takes 8 months to create a product after detecting a trend. With AI, we divide this time by three", concludes Frédéric Rose.
At the service of matter
Crédit photos : ©Textil'IA">At the upstream end of the fashion chain, the textile industry has also seized on AI. "Right from the start, we met with AI experts and listed possible applications to create innovative solutions with added value but accessible to our textile SMEs. The profession invests in these solutions via Textil'IA before deploying them. Contrary to popular belief, the cost of AI is not exorbitant, and it pays for itself fairly quickly," explains Robby Dubus of Textil'IA.
Textil'IA has already developed two solutions incorporating AI. Since 2023, its intelligent online product catalog has enabled Textilians to digitally archive their many designs: "we digitize their fabrics and transfer them to the platform equipped with AI-powered search tools," explains Robby Dubus. Users can easily search through their sometimes vast textile archives (several tens of thousands of patterns), via keywords or by similarity with another image, without having to manually categorize each pattern when importing it. "This intelligent catalog allows you to add value to your designs. It also facilitates exchanges between manufacturers and brands, by providing prospects with digital access to collections. Generative AI will also make it possible to analyze their searches and either make new graphic proposals or modify textile patterns to suit their needs."
The second AI-based solution signed Textil'IA facilitating anti-counterfeiting watch is called LOCKEO Détection. It enables Textilians to perform image recognition by comparing their designs with those present on online textile items. "For an SME, it would not be possible to hire someone to carry out such a watch seven days a week without even achieving these results."
Digital transition
Photo credit : ©Textil'IA">The third tool developed by Textil'IA -ODITH- already tested by some forty companies in the textile-clothing sector (including garment manufacturers) and marketed from this September, does not yet work with AI. But that doesn't rule out its use one day.
The aim is to facilitate traceability by securing the recovery, storage and transmission of manufacturing information for textile products. "Today, we're in the declarative business of suppliers. With AI, thanks to the cross-referencing of data, we could verify the provenance of a fabric, of a manufacturing origin, which would enable us to improve the solution" indicates Robby Dubus. Generally speaking, he emphasizes that "AI leads to solutions that facilitate the digital transition of companies. Without AI, it was impossible to digitize the entire creative heritage of textile companies!"
While he doesn't specialize in garment making, Robby Dubus imagines "surely plenty of AI applications, with tasks (such as patternmaking)" likely "to be largely optimized thanks to AI".
Automation and precision
Crédit photos : ©Haase Innovation">The experience of Haase Innovation shows that AI can indeed have a significant impact in garment making. Marine Anton, head of the workshop, recently invested in an automatic cutting machine managing patterns placed with on-board camera using AI. "It's very surprising, the camera will scan the pattern, we'll place the patternmaking on the computer and the machine will ensure that the patterns are connected between the back and the front," she explains.
The time saving is impressive: whereas it used to take employees 17 minutes to cut, by hand, a t-shirt with a placed pattern, it takes just 2 minutes 50 on the machine.
Marine Anton explains that her initial intention was not "to buy the machine to go faster" but to overcome a recruitment difficulty. "As the people doing the cutting would soon be retiring, we tried to replace them with young people". But this "physical, standing, long and repetitive work" attracted few.
This served as the trigger to invest in the automatic cutting machine and take the first steps into AI. It's a choice the manager clearly doesn't regret, even if it represents "without doubt the biggest investment ever made" in the workshop set up in 2014.
AI is gradually being integrated into the entire product process. Although it is already being implemented in the Design and material sourcing phases, its integration into garment making, where the intelligence of the hand is crucial, remains complex. However, it can improve competitiveness in manufacturing stages that do not require human added value.
For more information:
