Segment Anything (SAM)
First page
Paper summary
Meta's foundational model for image segmentation with massive training data release.
Ask this paper
01
Largest segmentation dataset: Releases SA-1B with over 1 billion masks on 11 million licensed images - by far the largest segmentation dataset ever.
02
Promptable segmentation: Introduces a new promptable segmentation task where users provide clicks, boxes, or text to indicate what to segment.
03
Zero-shot SOTA: Zero-shot performance is competitive with or superior to fully supervised specialist models.
04
Vision foundation model: One of the highest-impact vision papers of 2023, transforming how the field thinks about foundation models for dense prediction.