Computer vision used for fashion cataloguing

A visual search company has released a new platform that could save fashion retailers significant time, by automatically tagging images

Spotted: Bangalore-based fashion visual search company Streamoid has launched a new platform that uses intuitive AI, machine learning and computer vision to help retailers add product tags to images. The new product, AutoScribe, joins Streamoid’s stable of tools for visual search, web optimisation and personalisation — all aimed at fashion retailers. 

AutoScribe allows retailers to automatically catalogue new items and images, in what was previously a tedious manual process. While it may not seem like a big issue, large retailers might add hundreds or thousands of new products every month, making cataloging a challenge. The AutoScribe tool replaces manual tagging, and when the tags cannot be predicted confidently, AutoScribe sets it aside for manual quality control by stylists.

The software is designed to avoid issues specific to the fashion industry, such as the fact that a particular style can have many different descriptive names. A long-sleeved length might be called “long sleeved” in one marketplace, and “full sleeved” in another. This type of issue makes manual cataloguing a difficult process, which often requires significant expertise and training. AutoScribe allows the retailer to configure the taxonomy themselves, allowing them to control the nomenclature used. 

The product is already being used by Indian fashion powerhouse, Aditya Birla Fashion and Retail Limited (ABFRL). According to Praveen Shrikhande, Chief Digital and Information Officer at ABFRL, the company is “well aware that rich and accurate product information creates great customer experience online…With Streamoid’s AutoScribe, we now catalog our products in less than 6 hours instead of 2 days and our data is enriched immensely while maintaining an accuracy of 99 percent.”

The use of computer vision is proving to be a natural fit for fashion, helping online retailers to make real time outfit recommendations, and enabling users to quickly find products that they have seen in real life or online. Recent innovations in this field covered by Springwise have included incorporation of a body scanning app to create the perfect fit, and a shoppable streaming platform where users can buy what they see.

Explore more: Fashion & Beauty Innovations | Retail Innovations