Meta Launched AI Tool that Separate, Identify Items in Pictures

With its Metaverse aspirations wrecked, Meta is currently focusing on artificial intelligence to drive its next progressive phase. The Segment Anything Model is one of Meta’s most recent projects, the social media giant announced.

With just a few clicks, Segment Anything enables users to identify specific items in an image. The company claims that segment anything can already take a photo and individually identify the pixels that make up everything in the image while it is still in demo mode, allowing one or more items to be separated from the rest of the image.

Meta wrote in a post announcing the new model that segmentation, or determining which image pixels belong to an object, is a core task in computer vision. Meta stated that technical experts with access to AI training infrastructure and large volumes of carefully annotated in-domain data are required to create an accurate segmentation model for specific tasks.

By collecting a brand-new dataset of a previously unheard-of size, they outperform previous methods in terms of generalizability. Ross Girshick, a research scientist at Meta, added that, crucially, they did not limit the kinds of objects they annotated in this dataset.