Facebook launches what it identifies a “universal model of product recognition” that uses artificial intelligence to identify consumer goods, from furniture to fast fashion to fast automobiles.
It is the first step into a world in which the goods can be found and theoretically shopped for in any picture on the web. “We want to make anything and everything shoppable on the web, anytime the experience feels perfect,” Manohar Paluri, Facebook’s head of Applied Computer Vision told The Verge. “It’s a very fantastic idea.”
Product recognition is the first in a slew of AI-powered updates coming soon to its e-commerce platforms, the company says. Eventually, to create what it calls a “social-first” shopping experience, these will combine AI, augmented reality, and even digital assistants. Moreover, a tool called Shops has also been introduced today, enabling local companies to set up free storefronts on Facebook and Instagram. But as suggested by the Clueless reference, those features aren’t exactly new ideas. They were tried and tested even outside of Hollywood for years, often with mixed results.
Also Read: Electric Receives Another $7 Million in Funding From 01 Consultants and the Slack Fund
Amazon has also designed its own, now little heard-of, AI-powered fashion assistant with the Echo Look. And since the Amazon Fire Phone at least has become a practice using machine vision to recognize and search for products. Meanwhile, online retail sites such as eBay are now using AI to speed up the collection of products for sale and Amazon is one of a number of companies using machine learning that has launched its own “Shazam for clothes.” An AI-style assistant from Facebook’s dream could look something like this.
Facebook is saying its reach and precision are what make its software distinctive. GrokNet, the latest product identification platform created by the company, can recognize tens of thousands of different attributes in an image. Things vary from basic markings to color and scale stuff.
GrokNet has also been introduced on Facebook Marketplace, where it lets users list products for sale easily by recognizing what’s in them and creating brief descriptions. For eg, you could upload a picture of your couch, and Marketplace would recommend that you label it as a “black, leather, sectional sofa.”
The organization is also developing a version of this method which is designed for businesses. The AI program will automatically tag them and connect to shopping sites when they upload images to their account that contain their own items.
Facebook is supported by its exposure to images of users on the Marketplace in creating these devices. GrokNet is trained on a colossal database of around 100 million images in order of magnitude, with the majority taken from Marketplace. Facebook says this data is crucial for building a machine vision program that can recognize items in daunting lighting and from dodgy angles — a part of the online shopping experience that isn’t going anywhere.
Conclusion:
However, it’s unknown exactly how reliable GrokNet is. The organization claims it can classify 90 percent of retail photos in the segment Home and Garden, although it did not include comparable figures for other product segments.
