There has been much talk this year of voice search, yet not so much attention has been given to visual search, which has been quietly bubbling beneath the surface. There’s no denying that our world has evolved into a highly visual one, driven by the omnipotence of social media…so why shouldn’t we be able to search for what we need via images, as well as voice?
Several retailers have been making their foray in visual search this year, and this week Ebay became the latest to unveil its offering. In a nutshell, visual search allows machines to process and analyse imagery on a pixel-by-pixel basis, rather than leaning on human-defined search terms, such as keywords and descriptions. The most literate person in the world can sometimes struggle to find the right words to describe something they’re searching for, so it makes sense that search is moving forwards in this direction.
Within its mobile apps, Ebay is enabling users to “use pictures instead of words to search [its] 1.1 billion listings.” There are two strands to the initiative. The first, Find It On eBay, allows users to share an image they find on the web or through a social platform, such as Facebook or Pinterest, with eBay. Ebay will then find listings that are similar. So for example, a user could zoom in on a t-shirt they like in a blog post, to try and find ones that are similar. The second feature, Image Search, does pretty much the same thing using images that a user has taken and stored in their phone’s camera roll.
Ebay explains: “When you upload images to run Find It On eBay and Image Search, we use a deep learning model called a convolutional neural network to process the images. The output of the model gives us a representation of your image that we can use to compare to the images of the live listings on eBay. Then, we rank the items based on visual similarity and use our open-source Kubernetes platform to quickly bring these results to you, wherever you are in the world.” According to Ebay, this is just the beginning, and the technology has the ability to learn and refine itself over time.
See how it works within this Ebay video:
[iframe src=”https://player.vimeo.com/video/226972601?rel=0″ width=”560″ height=”315″]
Google and Pinterest are two businesses leading the way in visual search. Pinterest, for example, says it “can identify colours, shapes, textures…[and is] able to understand the combined affect people find appealing, even when it can’t be communicated in words”. Google unveiled its new visual search technology, Google Lens, earlier this year, which has been integrated into Google Photos and Assistant and can be used to help users identify what’s in their photos and videos and connect them to relevant resources, in real-time. One promising usage is identifying storefronts and other key business details when visiting a new area.
Stock photo company Shutterstock recently debuted a new “spatially aware” visual search tool that lets users find photos by composition, suggesting a slightly alternative use for the technology. Meanwhile for years, Facebook has been able to identify individuals in photos, which is some ways represents the holy grail for businesses wishing to make use of visual search. However the Facebook technology is limiting since it has only been set-up to identify faces, providing it with a clear target.
The phrase ‘an image can tell a thousand words’ may be a bit of a cliché, but there’s a lot of truth in it. Photos can offer far more context than words, showing exactly what the user is searching for, and resulting in much better refined search results. For any online business, speeding up the search process and delivering more accurate discovery and results can only be a good thing, to help increase sales, customer satisfaction and loyalty.
It’s still early days, and currently there is much awareness to be raised around visual search and how it works. Crucially, it needs to be intuitive and in line with how individuals are already using visual platforms such as Instagram, but if it succeeds, it’s easy to see why the technology could be popular. In this mobile era, people are searching less because they’re spending more time in apps and less time in the browser. Artificial Intelligence (AI) has also come on leaps and bounds, and so it’s possible to believe that visual search could end up being more accurate than traditional text-based search. The opportunities for innovation are extremely exciting, for online retailers particularly. I personally think it could make my annual winter boot hunt a much less arduous task!
