Style is often easily visualized but difficult to put into words. Now, furniture and home goods retailer West Elm is helping those at a loss for words to find furniture customized to their style by using Pinterest. The West Elm Pinterest Style Finder allows users to link to a Pinterest board for inspiration, then, using object recognition and machine learning, generates recommendations based on the Pinned styles.
The Style Finder is built much like Pinterest Lens, a beta program that uses a smartphone camera to generate relevant pins, often with the same pattern, colors, and designs. Luke Chatelain, vice president of innovation at West Elm, explains that while Pinterest Lens is designed to find a specific item, color or pattern, the Style Finder doesn’t look for exact matches. While both programs are built with similar software, the Style Finder uses a Pinterest board to define the individual’s style, coming up with suggestions in a similar, but not exact, atheistic.
“You can just say, ‘This is what I like, but I don’t know what it is.’ Style is an idea,” Chatelain said. “Recognizing a style that doesn’t have a word for it, this tool can decipher that and create similar results.”
The idea for the West Elm Style Finder was actually born in the company’s physical stores, where at the start of design consultations, most customers started by sharing a Pinterest board for inspiration. “That Pinterest board can really set the tone for the type of products [the consultant] wants to showcase,” Chatelain said. “In reality, technology can bring that to everyone online.”
To use the programs, users copy and paste the URL of any public Pinterest board — which means the board can be that individual’s board or a link to another user’s board, as long as that board is public. After pasting the link, users select to see results from the living room, bedroom, dining room and kitchen, office, or outdoor space. With that information, the system takes a few seconds and then lists three products from each category for that room, from wall art to sofas.
The Style Finder is an artificially intelligent program — West Elm developers worked with the Clarifai image recognition software as a basis, then developed front-end software that uses the recognition program for identifying, then suggesting, different decor styles. “Image recognition has really reached a place where a brand that is selling beautiful lifestyle products can leverage that to impact customers in a meaningful way,” Chatelain said.
Because the program is a neural network, the system will continue to improve over time. For example, when a user adds a product to the online shopping cart, the program learns that the suggestion was a good one.
West Elm will also continue to take customer feedback and integrate it into the program. The software is also helping the company improve its product lines by discovering gaps in available styles and pinpointing which styles are most popular.
The Style Finder, which officially launched on July 6, is designed to help users search even when they cannot come up with the right words to match the style they are looking for. The program is fluid, Chatelain explains — if the Pinterest board has multiple styles, multiple styles will be included in the results. And if a certain style is repeated in more images than another, the program will pick up on those reiterations too, prioritizing suggestions within that atheistic.
The West Elm Pinterest Style Finder is accessible from the company’s web page as an internet-based software.