May 18, 2018

Nextatlas Uses Clarifai's AI to Detect Maternity Fashion Trends

Table of Contents:

When researcher and Royal Melbourne Institute of Technology PhD candidate Anna Anisimova decided to do a research project on maternity fashion, she knew who to call. Anisimova wanted to trace the cultural, social and industrial currents and trends in maternity wear and Europe-based iCoolhunt’s trend research platform, Nextatlas, was just the ticket. Built with Clarifai’s computer-vision technology behind it, the trend detection engine can analyze content from insiders across several industries and detect trends. Most recently, Nextatlas combined their knowledge of trends and access to tens of thousands of user-labeled images in fashion, with Clarifai’s computer vision algorithm, to create and train a joint style model for this project and beyond!

Maternity Dresses built with Clarifai Computer Vision

For the project, they started with an archive of 1,035 images (retrieved from The Australian’s online national archives), including staged photos like magazine covers and spreads to paparazzi shots, then arranged the images by date (from the 1950s to the 2010s). After identifying and naming seven styles: “90s now”, “Boho”, “Dapper”, “Edgy”, “Minimal”, “Remix”, “Workwear chic” and eight types of fabrics and patterns (“Brocade”, “Camo”, “Denim”, “Eyelet and lace”, “Gingham”, “Polka Dot”, “Stripes”, “Tartan”), the goal was to identify when these were most popular in maternity fashion during the selected timeline. Every image was run through the model and labeled with the named styles, fabrics and patterns, each receiving a score out of 100. The results? Timeline curves that visually illustrated fashion trends over the years!

Texture and Fabric Detection Computer Vision Model

Thanks to the model, Anisimova knew “Boho”, for example, first peaked at the end of the 1960s and 70s, making a comeback in the mid-90s and again around 2010. “Its features clearly correlate with social and political movements,” Garella explained. “From anti-establishment to hipster.” “Tartan” fabric, meanwhile, had its highest peak in popularity in the mid-50s, then experienced a steep decline. While multiple spikes in popularity have occurred over the years, its presence in maternity fashion largely flatlined since the mid-90s.

The full results will be available in Anisimova’s final research, but Nextatlas sees endless possible applications for the model. “Given a larger sample and custom style models, as deemed relevant by market or industry or cultural movements,” Garella said. “The results would coincide with the creation of sector overviews and offer fertile ground for novel and creative market insights.” This research also offers many opportunities to build strategies on top of brand heritage. As Garella explained, “visual history has overcome written texts in archives and offers itself as an identity storage waiting to be used and better understood.” Nextatlas continues to work on ideas to further improve the detection of their style models and we look forward to seeing how next they apply our technology!

“We’ve been working with Clarifai for some time now and it has been a very interesting and fruitful collaboration.” – Luigi Garella, Nextatlas Insights Manager

Ready to get started? Get in touch. 

How to Drive Revenue with Computer Vision AI