Artificial Intelligence that Gives Fashion Advice Now Developed
Wed, April 21, 2021

Artificial Intelligence that Gives Fashion Advice Now Developed

The market value of the fashion industry in the US alone is 406 billion dollars. / Photo by: S_L via Shutterstock

 

A computer science team from the University of Texas at Austin has successfully developed an artificial intelligence system that can look at a picture of clothes and suggest fashion styles to make it more fashionable, reports journal Science Daily.

The “robot” wears Prada

In collaboration with researchers from Facebook AI Research Lab, Georgia Tech, and Cornell Tech, the Texas team has created an AI tool that can give suggestions. Tweaks include selecting a longer jacket or a sleeveless top. Computer science professor Kristen Grauman, whose past research was focused on AI visual recognition, said that they thought of the system as a friend giving fashion advice or feedback to another person.

Fashion++

The team named their work Fashion++. It utilizes visual recognition systems to help analyze the shape, texture, pattern, and color of the garments in the photo. Then, it analyzes where changes are necessary and creates the most impact on the overall look. Then, it offers another alternative for the users to see.

Professor Grauman and the team trained the artificial intelligence tool to use over 10,000 images of clothes that have been shared publicly through the internet. Most of these images were from fashion enthusiasts. Graduate student Kimberly Hsiao shared that finding those images was relatively easy. What was challenging to them was finding the “unfashionable” ones. To overcome the challenge, she combined both the fashionable images and the unfashionable ones so that the system would be trained to work with clothes that are not good to wear.

Hsiao pointed out that it would be easy for the AI to learn as fashion also evolve. It simply requires adding new photos, which there are plenty of online. The computer science group will be presenting their work at the International Conference on Computer Vision in South Korea.

Limitations on the AI

As of now, the team has recognized that their artificial intelligence may be biased at some point since vintage looks are difficult to consider as stylish because the images were more trained with modern fashion images. Furthermore, most of the images were from North America as styles from other countries don’t usually show up on the internet. They also acknowledge the fact that the majority of the images were worn by models and bodies appear in different sizes and shapes, so it may affect the fashion choices of their technology. To make the recommendations more tailored, the team continues to work on letting Fashion++ learn of things that flatter other body shapes.

Fashion industry and AI

The market value of the fashion industry in the US alone is 406 billion dollars. An estimated 1.8 million people in the US are also employed in the fashion industry, including 232,000 people manufacturing textiles for apparel as well as other fashion items. This is according to the Bureau of Labor. One of the largest fashion companies, Nike is also American and generates an annual revenue of 30.6 billion and about 105 billion dollars in market value. There is no denying that the fashion industry is indeed one of the biggest globally among other industries. In 2018, the fashion industry was estimated at 3 trillion dollars, said AI-focused research and advisory firm Cognilytica’s managing partner and principal analyst Ron Schmelzer.

Apparel and clothing brands are also constantly looking for new ways to get their products in front of their buyers to create market demand and awareness. This paved the way for the use of machine learning and artificial intelligence. So far, AI has helped maximize the shopping experience of users and improve the sales system’s efficiency.

Schmelzer believes that because of the constant changes in design and fashion, retailers also need to keep up with the trends and predict the preferences of consumers for the next season. Using AI can reduce the forecasting error by up to 50 percent. 

 

AI has helped maximize the shopping experience of users and improve the sales system’s efficiency. / Photo by: oneinchpunch via Shutterstock

 

The E-commerce fashion industry, artificial intelligence: statistics

The Fashion and Apparel Industry report shows that worldwide revenue of the e-commerce fashion industry will increase from $481.2 billion the previous year to $712.9 billion by 2022. This year, it will be worth $545 billion, will increase to $606 billion in 2020, and $662 billion a year after. Factors that will drive such growth are the expanding worldwide markets outside the West, smartphone penetration and increasing online access, and innovative tech that creates experiential online shopping.

E-commerce personalization platform Nosto also shared that personalization is the leading factor in e-commerce. Forty-three percent of purchases in the fashion industry are influenced by personalized promotions and recommendations. Then, 74 percent of consumers choose brands that have personalized messaging, experiences, and offers. About 94 percent of firms also consider personalization as “critical” to their future and current success in the industry. A separate study from MIT has, however, revealed that only 23 percent of businesses incorporated artificial intelligence into their product or service offerings today and the most likely to have an AI strategy in their businesses are the largest companies or those with at least 100,000 workers.