HYPEBEAST - Meta has developed a new AI model called the Segment Anything Model (SAM) which can identify specific objects in images and videos, even those it hasn't encountered before. It has also created the largest dataset of image annotations to date, called the Segment Anything 1-Billion mask dataset (SA-1B), using SAM's functions. SAM can develop masks for any object in any image or video and is versatile enough to include a broad set of image "domains."
by Dylan Kelly
See full article at Hypebeast