Wayfair has launched a new visual search tool for their product line that is powered by a machine learning AI built upon trends, attributes, and design tropes from their own massive catalog of furniture and home decor. That catalog allowed Wayfair to train up a very specially-made AI for its specific niche, being to search Wayfair's products for visually similar offerings to what a user feeds the visual search tool. Because of this, and how specialized it is, Wayfair's tool does its job better than most visual search tools out there, including Google's.
The way it works on the user side is fairly simple and intuitive. Whether on the app or on the web, all a user has to do to kick off the process is tap the camera icon to feed Wayfair a picture. This can be a freshly taken photo from the app, or something on a user's device in either environment. From there, the user can then select the part of that picture that contains the item that they want to search for. Wayfair will then pass that imagery on to the AI, which will break it down and figure out the various design tropes and attributes present. It does this by visually examining the image against its wide database, looking for visual cues like color, item type, specific design conventions, or overall design styles, such as nature-inspired, Art Deco, or futuristic.
Once the AI has deciphered the imagery presented to it and come up with a suitable list of traits, it will look through Wayfair's catalog for items that check a number of those boxes, then present those to the user. This results in the user not only being able to hunt down a specific item on Wayfair if they see it around town or in a friend's home f0r example, but also allows them to snap a picture of something in their own home and be shown items that could go well with it and help tie the look together. Items found via this search can be bought immediately, or sent to an Idea Board to be paired up with other items for a complete set.