By AI Trends Staff
The fashion industry did $3 trillion in business, 2% of global GDP in 2018; e-commerce fashion amounted to $520 billion in 2019. It’s a big business. AI is poised to revolutionize the fashion industry by providing insights into fashion trends, purchase patterns, and enabling better inventory management.
Capgemini, the French consulting services multinational, estimated that global annual spending on AI by retailers is projected to hit $7.3 billion by 2020, and AI could help retailers save a potential $340 billion annually by optimizing processes and operations, according to a recent account in T_HQ technology and business.
The global brand H&M has been applying AI solutions to boost business operations. One example is a system to organize and allocate masses of unsold stock to retail stories with highest demand, reducing the need for discounted sales. This is achieved by optimizing the supply chain and inventory management, reducing the amount of wasted clothing.
The fashion industry continues to be among the biggest global polluters, responsible for 10% of global carbon dioxide emissions, 20% of the world’s industrial wastewater, and 25% of all insecticides used in the industry, according to an account in TowardsDataScience. AI can be used at many stages of production to address the pollution issues and improve working conditions. The industry envisions using machine learning, deep learning, natural language processing, visual recognition and data analytics to reduce errors in trend predictions and perform more accurate forecast trends.
UK online fashion company Asos launched an AI-powered tool to help shoppers find the perfect fit, a notorious flaw in online shopping as customers are unable to try on a garment before purchase. The new AI sizing tool, called Fit Assistance, provides a recommendation after asking shoppers a list of questions such as age, height, weight, and body measurements.
“This recommendation is based on the size that people like you bought, and whether they returned it,” as stated on the company website. Furthermore, Fit Assistance also reveals the percentage of shoppers that were satisfied with the recommended size.
Tommy Hilfiger worked with IBM and the Fashion Institute of Technology in New York on the “Reimagine Retail” project in 2018, to equip fashion designers with AI skills for designing. According to Steve Laughlin, the general manager of IBM Global Consumer Industries, their aim was to speed up the supply chain process and aid the next generation of retailers with AI powered skills, according to an account in Analytics Vidhya.
IBM gave access to their AI facilities for the FIT students for this project; including access to their natural language understanding and computer vision labs as well as several deep learning techniques trained specially with fashion data.
All these tools were then applied to 15,000 Tommy Hilfiger’s product images, along with approximately 600,000 publicly available images, taken from various fashion shows. Close to 100,000 patterns were taken from fabric sites. The resulting model then churned out tons of patterns, trends, silhouettes and prints that enabled the FIT students to create completely new designs by incorporating the trends of other designs into the ones already existing in the Tommy Hilfiger database.
The ‘Reimagine Retail’ project also uses social media listening as a tool to understand how previous products have been received and make changes in upcoming designs. Predicting which items are going to be in style in the coming months and years has become critical for retailers.
“The goal was to equip the next generation of retail leaders with new skills and bring informed inspiration to their designs with the help of AI,” stated Avery Baker, the chief brand officer at Tommy Hilfiger in an IBM blog post at the time. (She has since left the company.) “AI can identify upcoming trends faster than industry insiders to enhance the design process.”
Among the final designs created by the students was a plaid tech jacket made with advanced color-changing fibers that responds to AI analysis of voice and social media feeds. The design was to use eco-friendly materials and demonstrate what the future of product customization can be.
“As a brand, we are always pushing the boundaries of what’s possible through innovation and disruption. These young designers truly embody this spirit by showcasing the successful integration of fashion, technology, and science,” Baker stated.
FIT students involved in the Reimagine Retail project, all FIT fashion design majors, published observations in an account in Business Insider.
“As a fashion designer, I tend to stay in my own head, but with these tools, I was able to look into databases that were curated with an incredible amount of information, which, in turn, inspired in new ways that I could make design decisions faster,” stated student Grace McCarty.
“The tools that I used were the silhouette recognition tool, the color analysis tool, and the print tool, which make the designer’s job easier and more efficient but does not take their role. AI technologies and fashion designers will be in a symbiotic relationship,” stated student Amy Taehway Eun.
She added, “As a result of this project, I can see that the relationship between AI and designers will be collaborative. Technology will help designers create new and fresh products. In the end, I believe a designer’s role will be the same. They will research inspirations, they will design, but they will have better options and a different perspective offered to them by AI technology.”
New Effort Applies AI to Creative Side of Fashion Design
A newer effort is focusing on using AI to enhance creativity in fashion design, going beyond streamlining supply chains and managing inventory better.
“Initial uses of Artificial Intelligence have focused on quantifiable business needs, which has allowed for start-ups to offer a service to brands,” stated Matthew Drinkwater, head of the Fashion Innovation Agency at London College of Fashion, in a recent account in Forbes “Creativity is much more difficult to quantify and therefore more likely to follow behind.”
London College of Fashion recently launched an eight-week AI course for 20 volunteer fashion students to learn Python to write code to gather fashion data, then use it to develop creative fashion solutions and experiences.
The AI course was developed by the Fashion Innovation Agency (FIA) in partnership with Dr Pinar Yanardag of MIT Media Lab, and FIA’s 3D Designer, Costas Kazantzis.
The AI models used were generative adversarial networks (GANs), a type of machine learning where two adversarial models are trained simultaneously. The generator (“the designer”) learns to create images that look real, and a discriminator (“the design critic”) learns to tell real images apart from fakes. During training, the generator becomes better at creating images that look real, while the discriminator becomes better at detecting fakes. The application of this creatively allows computer-generated imagery and movement that look plausible (and likely aesthetically pleasing) to the viewer.
A pivotal output from the course was a virtual fashion show created from archive catwalk show footage, placed in a new 3D environment with the models wearing new 3D-generated outfits. Drinkwater stated that this is an example of “how even those with limited experience in the field can collaborate to push boundaries.”
Style transfer was used to apply image recognition software to recognize patterns, texture and colors, and then suggest designs and placement on the garment. The 3D environment for the virtual show was created in gaming engine Unity.
The proof-of-concept virtual show was to launch in September on the fifth day of London Fashion Week, which is operating in a decentralized manner across digital and physical platforms.
Now in its initial stages, these experiences show a promising future for AI in fashion design.