When Google Lens was introduced in 2017The research functionality has accomplished a feat that did not seem so long ago would have been science fiction: point the camera of your phone in an object and the Google lens can identify it, display a context, perhaps even let you buy it. It was a new way of looking for, which did not imply tapping descriptions of things you saw in front of you.
The objective also demonstrated how Google planned to use its automatic learning and AI tools To ensure that its search engine appears on each possible surface. As Google is increasingly using its fundamental generative AI models for generate information summaries In response to text research, the visual research of Google Lens has also evolved. And now, the company claims that Lens, which feeds around 20 billion research per month, will take care of even more ways to search, including video and multimodal research.
Another adjustment of the objective means even more context for purchases will appear in the results. Purchases are, not surprisingly, one of the main cases of use for the objective; Amazon and Pinterest also have visual search tools designed to supply more purchase. Look for your friend’s sneakers in the old Google goal, and you may have shown a carousel of similar items. In the updated version of the objective, Google claims that it will display more direct links for purchase, customer reviews, publishers and comparative purchasing tools.
With the kind permission of Google
The search for objectives is now multimodal, a warm word in AI these days, which means that people can now search with a combination of video, images and vocal entrances. Instead of pointing the camera of their smartphone on an object, pressing the point of focus on the screen and waiting for the lens application stimulates the results, users can point the lens and use Vocal commands at the same time, for example: “What type of clouds are these?” Or “what brand of sneakers are these and where can I buy them?”
The objective will also start to work on real-time video capture, by passing the tool beyond the identification of objects in fixed images. If you have a broken disc tray or see flashing lighting on a defective device at home, you can make a quick video via the lens and, thanks to an AI generator overview, consult advice on the way of repairing the element.
Announced for the first time at IS / S, this feature is considered experimental and is only available for people who have opted for Google’s research laboratories, explains Rajan Patel, an 18 -year -old Googler and a co -founder of Lens . The other features of Google Lens, vocal mode and widen purchases, take place more widely.
The “video understanding” functionality, as Google calls it, intrigues for several reasons. Although it currently works with videos captured in real time, if or when Google extends it to captured videos, whole video standards of videos – whether in a person’s own camera or in a database Gargantusensene like Google – could potentially become labeled and extremely buyer.
With the kind permission of Google
The second consideration is that this objective functionality Share certain characteristics with the Google Astra projectwhich should be available later this year. Astra, like the objective, uses multimodal entries to interpret the world around you via your phone. As part of an astra demo this spring, society has shown a pair of prototypes of smart glasses.
Separately, Meta has just caused a sensation with her long -term vision for our future of augmented reality, which just involves Mortals wearing silly glasses This can intelligently interpret the world around them and show them holographic interfaces. Google, of course, has already tried to achieve this future with Glass Google (which uses a technology fundamentally different from that of the latest Meta argument). The new features of Lens, associated with Astra, a natural sequence of a new type of intelligent glasses?