While Amazon works to accomplish a frictionless prospect in fashion, the organization is testing a way to produce new clothes without personal designers.
News reports that a crew at the e-commerce company, operating at a San Francisco research center, has devised an algorithm that examines images to learn about particular styles of clothing. It then produces similar new items from scratch. The technology is still in early steps and not ready for Amazon to put to use, but it shows what the organization might have a chance in the market for the future.
To do it, the organization uses an AI system called generative adversarial network (GAN), basically developed by a Google researcher. It works by holding two deep neural networks look at models of existing fashion, and then go back and forth with each other, like a designer-critic team, to build something new that meets certain criteria. A shirt, for example, must have two sleeves. It learns from past human creativity in order and, in a thought, replicates it.
It’s not completely clear how Amazon would put this technology to use. Apparently, it could examine what people are purchasing and see where there are gaps in the market, using the new AI to generate ideas for its many private fashion labels that could fill those gaps. Stitch Fix, a subscription clothing service, is currently using AI for that idea and says the sales have been great.
Stitch Fix does, nevertheless, use human designers who turn the AI’s plans into an actual garment. Whether Amazon’s technology would need that kind of direction is uncertain, but if not, it raises the question: Could the algorithm tell if a look it makes is a dud? An Amazon spokesperson said the organization has no further information it can provide.
It’s maybe telling that Amazon has also been going on AI that can judge whether an equipment is stylish. It recently published Echo Look, which lets users take photos of their outfits using a camera powered by Alexa, the company’s interactive voice service. Among its characteristics is the option for users to get a second view on an outfit they’ve recorded, using machine-learning algorithms coupled with advice from fashion experts at the company.
Take your time to comment on this article.