After the collapse of its Metaverse ambitions, Meta (formerly known as Facebook) announced its latest project called the Segment Anything Model.
This new AI tool allows users to identify specific items in an image with just a few clicks. While still in demo mode, Meta says that Segment Anything can already identify the pixels that makeup everything in a photo, allowing one or more items to be separated from the rest of the image.
According to Meta, segmentation, or the process of identifying which image pixels belong to an object, is a core task in computer vision and is used in many applications such as analyzing scientific imagery and editing photos. To create an accurate segmentation model for specific tasks, technical experts require access to AI training infrastructure and large volumes of carefully annotated in-domain data.
Meta’s Segment Anything AI system was trained on over 11 million images and is now available for research under a permissive open license.
The company claims that its model shows impressive capabilities for handling various types of images, including ego-centric images, microscopy, or underwater photos, thanks to the scale and generality of the data.
Although the tool is still in its research phase, there are already concerns about potential privacy law violations in the use of facial recognition without express consent.
In February, Meta pivoted from its plans to launch a metaverse and announced its shift to focus on other products, including artificial intelligence, after laying off over 10,000 workers following the termination of its Instagram NFT project.
Generative artificial intelligence is an AI system that generates text, images, or other media in response to prompts. Meta’s focus on AI and its latest project, Segment Anything, marks its commitment to driving its next stage of development through AI.
