Meta Releases SAM 3 and SAM 3D to Reconstruct the Physical World
New open-source models allow users to segment and build 3D objects from simple text prompts.
Meta AI advanced the field of computer vision, with the release of SAM 3 (Segment Anything Model 3) and SAM 3D. The models push beyond 2D image recognition, offering developers the ability to reconstruct three-dimensional objects and human bodies from single flat images.
SAM 3: Promptable Segmentation
The third iteration of the Segment Anything Model introduces "Promptable Concept Segmentation." Unlike previous versions that required clicks or bounding boxes, SAM 3 can identify and segment objects based on abstract text descriptions (e.g., "yellow school bus") or image exemplars. * Benchmark: To validate performance, Meta released the SA-Co benchmark, featuring over …
Archive Access
This article is older than 24 hours. Create a free account to access our 7-day archive.