quick search of #fashion on Instagram brings up more than half a billion results. It would be nearly impossible for one person – or even a dedicated team – to tease out meaningful trends and insights from such an onslaught of visual data.

For an AI (properly trained with the right algorithms), it’s a piece of cake, according to Kavita Bala, chair of the computer science department at Cornell University. She and her team used artificial intelligence (AI) to create a map of style trends and influencers by analyzing 14.5 million photos of people shared publicly on social media. Bala’s StreetStyle project can answer questions like: How many people wear black in Los Angeles today, compared with two years ago? Or, where in the world is the hijab most prevalent?

“You can, in fact, see the cities or regions wearing particular styles and colors, which reveals something fundamental about how people use fashion to express themselves,” Bala says.

Though the StreetStyle project is not currently being used commercially, a number of retailers (including Amazon) are using AI to up their fashion game – whether it’s to cut production time, make personal style recommendations, sniff out potential copyright infringements or design new garments.

“Artificial intelligence is about tapping the untapped potential,” says Bala.

Fashion brands that embrace AI are more agile and better equipped to thrive in the cutthroat apparel industry, experts say. Those that ignore the tech do so at their own peril. “[AI] is providing more accurate information for better decision making,” says Paulo Sampaio, a data scientist at EDITED, a tech company that helps brands like Topshop and Ralph Lauren. “We all know retail is extremely competitive, and it’s even harder to manually monitor everything that’s going on. AI provides the information you need to keep your competitors in sight and monitor new players you didn’t even know existed.”

efore delving any further, it’s important to understand what AI is and is not. It’s not the self-aware – and often murderous – entities popularized in Hollywood blockbusters. Neither is it things like sewbots or mechanical robot arms; those fall under the realm of automation – a set of strict preprogrammed rules with no ability to interpret data. Instead, AI is a broader field that deals with emulating the human mind and reasoning. It focuses on training machines to use algorithms that discover patterns and generate insights and make predictions based on data already out there.

As the reigning monarch of e-commerce, it makes sense that Amazon would be at the forefront of fashion-forward AI. The retail giant has been making a push into online apparel sales – beyond basics like socks and undershirts – in recent years. Amazon researchers based in Israel developed AI that can analyze labels attached to images and determine whether a particular look is stylish, according to an article in MIT Technology Review. The company’s team at Lab 126, a San Francisco-based research center, developed an algorithm similar to Bala’s StreetStyle project. Amazon’s rudimentary AI fashion designer “learns about a particular style of fashion from images, and can generate new items in similar styles from scratch,” according to the MIT article.

Podcast

Sarah Krasley, CEO of Shimmy Technologies, shares insights into how artificial intelligence will help apparel companies stay competitive.

But Amazon isn’t alone in analyzing fashion trends using AI. EDITED uses AI to help major retail brands spot trends, understand consumer preferences, analyze price and stock movements, and produce new variations of garment designs.

“We use artificial intelligence to identify, classify and organize millions of products,” Sampaio says. “That way, a designer can easily find specific products that, for instance, are selling fast, sold out or high-priced. Designers even have insights into color wheels and sizes within very specific product segments.”

A footwear designer could ask EDITED’s AI to examine whether heeled espadrilles, for example, are on-trend and which colors and shapes are most popular, then use that information to redesign or improve their product.

ecorated-apparel companies are also delving into the world of AI, though not necessarily to generate new designs or identify upcoming fashion trends. Scalable Press (asi/87178), which does significant business in the print-on-demand space, uses AI to weed out copyright infringement.

Initially, Scalable Press hired a team of moderators to monitor user-uploaded content for potential problems. They would examine incoming artwork and make a decision about whether the image contained a trademark or copyright that the user didn’t have the rights to, says Eric Zhang, vice president of engineering.

“Very quickly, we realized human moderators don’t work,” he adds. “They’re too expensive; plus, humans make mistakes. Computers don’t make the mistakes humans do. A computer’s mistakes are really just humans not programming well enough.”

Scalable Press used a combination of techniques to create its AI. “We compare images with good and disallowed images, we look at the text, we look at the sellers themselves and whether there are any risk factors,” Zhang says. “We put all of these factors into a statistical model, and it tells us if this should be approved, rejected or needs human moderation.”

The AI isn’t 100% accurate – there are false negatives and false positives – but Zhang estimates that about 90% of Scalable Press’s moderation is done via AI without the need for human intervention. Over time, Zhang adds, that percentage will improve. “When we take anything down, we add it to our database,” he explains. “Every image is training our system to be better. … Whenever a decision is made, it’s used to train our systems to become better at screening.”

Other companies are using AI to streamline their workflow. Shimmy Technologies, a Brooklyn apparel and tech startup, has incorporated artificial intelligence to help speed up the apparel design process. Founder and CEO Sarah Krasley was inspired to explore technology when she launched a swimwear line, after watching an automotive designer sculpt the hood of a car and wondering whether the same techniques could be used for human bodies.

“Artificial intelligence is about tapping the untapped potential.”Kavita Bala, Cornell University

Swimwear is notoriously difficult to measure, according to Krasley, and tailors typically have to put their measuring tape down multiple times to make notes. To streamline the process, Krasley partnered with IBM’s Watson AI: Tailors simply had to say a measurement, and Watson would plug it into the computer to create a 3-D model of a design. Using AI cut Shimmy designers’ worktime by 20%, according to Krasley.

Designing swimwear was just the beginning for Krasley. She’s since morphed Shimmy Technologies into a cloud-based platform that uses AI to improve workflows and help brands increase design and manufacturing efficiencies. Since the company is still in its beginning stages, Krasley says she’s not able to disclose specific details, but hinted that future plans will transform the apparel sector. Shimmy Technologies is also in the process of launching an application called Shimmy Upskill, a new approach to help workers downsized due to automation and AI. The application is an attempt to shift the scary narrative away from robots stealing people’s jobs to one of AI being the “great skills enabler” in the workplace. In fact, an Accenture report found that AI is projected to increase labor productivity by up to 40%, which will enable workers to make more efficient use of their time.

I has been a valuable tool for companies making connections with end-users. Thanks to technology, almost anyone can reap the benefits of a personal stylist – as long as they’re not insistent that said stylist has a heartbeat.

You’ve probably heard of Stitch Fix, the online styling service company with $1 billion in revenue and 2.2 million active customers last year. The San Francisco-based company sends customers boxes of clothing, which they can decide to keep or return. The firm uses both human stylists and artificial intelligence to analyze style trends, body measurements, customer feedback and preferences to provide customers with recommendations.

Stitch Fix even uses something it calls “genetic algorithms” to design new apparel styles for its own brands. The company describes the process in exhaustive detail on its website, but in essence, it looks at user feedback for things like color, sleeve style or hem length and creates “new styles by recombining attributes from existing styles and possibly mutating them slightly.” These new styles are vetted and approved by the human stylists before they ever hit a client’s doorstep, though.

Another company concerned with personal style is Pureple, an app that uses AI to help people define their personal style by recommending outfits from their own closets. The Pureple app was developed by Nazan and Ben Kurt, a married couple who are both former Microsoft employees. The working parents of two believe people spend too much time getting ready, so they developed this virtual stylist app to help streamline morning routines.

Users can upload pictures of their apparel and accessories to create a digital wardrobe, and the app uses machine learning to harvest the information it knows about style and trends to help choose an outfit for the day.

“I’m a big fan of optimizing and bringing structured solutions to our daily life, and I also like fashion,” Nazan Kurt says. “But I also had an ongoing challenge with it, because I felt it shouldn’t be this hard to be stylish.”

Though many of the AI solutions being explored in the fashion world are still in their infancy, they’re only going to improve with time. It’s a disruptive technology likely to impact even the stereotypically set-in-its-ways decorated-apparel business. “Printing in particular is slower to adopt technology than other industries,”Zhang of Scalable Press says. “I would like to see everyone utilize technology. It’s about reducing waste and offering a better customer experience.”

RACHEL RAMIREZ is an editorial intern for Wearables.

Product Hub
Find the latest in quality products, must-know trends and fresh ideas for upcoming end-buyer campaigns.