Pinterest was created as a place to collect and search for interesting things, usually images and pictures, on the Internet. But not every interesting thing has to or can be found online. Sometimes inspiration strikes us when we see an object in the real world. For those occasions, Pinterest has been developing its Lens visual discovery feature, which it is now making available to everyone in the US, though still in its beta form.
Lens isn’t simply just a way to take a photo of an object and upload it to Pinterest. That is too easy and already a well-supported action. Instead, it is the visual counterpart of the text-based search that Pinterest already has, allowing you search for pins already on Pinterest that’s related to the objects in the camera’s viewfinder. What’s interesting is that Pinterest Lens can actually, or tries to, identify different objects in the image to make multiple suggestions.
It may all sound simplistic from an end user point of view, but from a computer science perspective, it’s quite a feat. Object recognition is still uncommon as far as end user features go, but that might change in the coming months. Pinterest already has a sort of visual recognition feature but that only applied so far to images already stored on Pinterest. Taking live images of real world objects and identifying them is definitely something commendable.
That said, Pinterest admit there’s still a lot to be done, which is why it isn’t shedding the beta tag yet. While Lens is now rolling out to Android and iOS for the public, it is limiting it to the US public. It is also seeking help from these users in refining Lens for when it makes mistakes. And it definitely will.