As everyone’s well aware, Facebook has been fairly blatantly copying Snapchat’s core features in an effort to stop their audience from migrating across to the ephemeral content app.
But Snapchat isn’t the only social platform Facebook’s been duplicating. Aside from their Snapchat-like tools, Facebook’s also added in features which are similar to Slack, LinkedIn, Houseparty – basically, Facebook has more capacity and more data than anyone else, and using that, they can effectively challenge any rising technology, should they choose to do so. Should that rising tech pose a threat.
Which is why this new discovery is of particular interest.
Spotted by Facebook user Jane Wong (and highlighted by The Next Web’s Matt Navarra), it appears that Facebook’s testing a new tool which would enable users to search objects in Facebook’s Marketplace via an image.
The option makes a lot of sense, and it clearly aligns with Facebook’s ongoing development of their image-recognition technology – but it also looks very much like this:
Pinterest launched their image-recognition centered search option ‘Lens’ in February, which enables users to take photos of anything and get a listing of Pin matches. And the option’s proven popular – Pinterest says Lens use tripled between April and May alone, while monthly visual searches on the platform have increased almost 60% over the last year.
It seems Facebook has also taken note of this popularity – though, really, visual search is the next evolution, and one which Facebook has been working towards for some time. Maybe, by testing it out within Marketplace, it gives Facebook a safer testing ground for the next evolution, which will see a full-scale visual search rollout.
As noted, Facebook’s been developing their image-recognition technology for some time. No doubt you’re aware of their facial recognition tools, which can pick out users in images, but more than that, Facebook’s been developing advanced computer vision technology that can pick out not only specific objects in a frame, but can also identify actions and movement.
There’s a range of ways in which Facebook could utilize this – already, Facebook uses their image-recognition AI to identify and remove objectionable content (in 2015, their machine learning models examined and eliminated more of these images than people did). Advanced image recognition could give the platform the upper hand in a range of search functions, like related product searches, and given that there’s somewhere in the vicinity of two billion photos being uploaded to Facebook, Instagram, Messenger and WhatsApp every day, their potential research pool is much larger than that of their competitors, giving Zuck and Co a distinct advantage.
So should Pinterest be worried? It depends on how Facebook uses it. Pinterest’s Lens is very product focused, which aligns with platform usage. An expanded Facebook system along the same lines could definitely have some impact on Pinterest’s growth, but as noted, visual search has been coming for a long time anyway. Pinterest just got the jump by releasing Lens earlier than other platforms.
Of course, Google has Google Images, and there are similar product options being built into those searches too, but the direct eCommerce focus of Pinterest makes Lens a little different.
Facebook, at least in this test, looks to be coming after them on that front – though I’d expect Facebook to also eventually tie their image recognition tech into their evolving AR capacity, enabling businesses to provide more advanced ads and insights based on the recognized object.
Really, this seems like a relatively small test for Facebook, another way to boost their image recognition capacity and get a better handle on potential use-cases and interest. And in Marketplace, it makes a lot of sense – you might be looking for a specific item to replace one you’ve broken, a specific tool that you don’t know the name of. This system could help find them – while also helping fuel Facebook’s image recognition capacity ahead of their next big push.