Updated: February 24, 2019 11:12 PM
A Duluth fashion content agency had a big role in the Oscars red carpet on Sunday. Runway Manhattan launched its website, FashionBrain, an artificial intelligence technology that provides detailed information in seconds on what celebrities are wearing on the red carpet.
"It is a pretty bold move that we say we are going out, a little company in Duluth, with the technology we have created together with BrainCreators, and take a stab at the world’s largest red carpet and let it tag that red carpet automatically,” Markus Muller, the editorial director and co-owner of Runway Manhattan, said.
Runway Manhattan teamed up with BrainCreators, a European artificial intelligence company to launch a high-tech tool that processes fashion images and instantly assigns keywords to them, with fashion terms related to apparel, hair, makeup, jewelry, and other details.
Runway Manhattan said they started out in 2014 with their Duluth, Minnesota-based team manually tagging fashion imagery in an effort to make fashion photos as accessible and easily discoverable as possible for their clients. They collaborated with BrainCreators in 2016 to develop FashionBrain.
“A lot of the concepts that a fashion editor looks at and fashion editors would look for in imagery are actually detected properly, accurately, and detected within milliseconds. The system will be able to pass through hundreds and thousands of images and display a result that will help make stories about the Oscars,” Muller said.
“Runway Manhattan’s recent collaboration with BrainCreators leveraged a huge dataset of annotated fashion photography with state-of-the-art artificial intelligence technologies to automatically tag fashion photos, using A.I. models that select from thousands of industry tags such as hair color, clothing style, and accessories,” BrainCreators explained in a company statement.
In a demonstration of how FashionBrain works, Muller took an image online of a celebrity posing at a red carpet and had it go through FashionBrain’s A.I.
The website detected the woman’s hair color, the style of her hair, the style of her dress, which was a full length strapless dress, and went into further detail by identifying the color of the dress and also by identifying the celebrity in the image.
The website is described as an online fashion detector that works with the click of a button.
"We are running something we have been thinking about for a couple of years and have been working on for about five years, so it's a culmination of a lot of work,” Lara Signorelli, the photography director and co-owner of Runway Manhattan, said.
The A.I. is trained by using Runway Manhattan’s extensive archive of over 500,000 images that are keyworded with thousands of fashion-related terms that are referred as tags.
Muller said those “perfectly” keyworded images on Runway Manhattan’s archive has given them the edge that makes FashionBrain efficient. They have compared it to similar website’s like Google Vision and Clarifie and said FashionBrain is superior to them in detail.
"Long term you can think about other photo providers being able to live tag imagery as the Oscars are going on, as the red carpet is going on," Signorelli said.
"FashionBrain is our first attempt to create systems that make it possible to accommodate fashion publishing of the future,” Muller said.
Runway Manhattan said this will help media specializing in fashion tremendously by tracking specific looks and trends instantly.
"They'll run their images through FashionBrain and they'll get perfectly keyworded images so they won’t have to scroll through thousands of image just to find the one they're looking for,” Zach Chase, the chief editor for Runway Manhattan, said.
"It will make it easier in an instant for fashion editors worldwide to create meaningful content for their many audiences using images that have been passed through FashionBrain,” Muller said.
A demo of the website can be found here.
Updated: February 24, 2019 11:12 PM
Created: February 24, 2019 05:16 PM
Copyright 2019 WDIO-TV LLC, a Hubbard Broadcasting Company. All rights reserved