Segment Anything

TL;DR

The paper presents the Segment Anything project, introducing a new image segmentation task, a large-scale dataset with over 1 billion masks, and a versatile promptable model that performs well in zero-shot settings across various tasks.

Contribution

It introduces the largest segmentation dataset to date and a promptable model capable of zero-shot transfer, advancing foundation models in computer vision.

Findings

Zero-shot performance is often competitive with or better than supervised methods.

The dataset contains over 1 billion masks on 11 million images.

The model demonstrates strong transferability across diverse segmentation tasks.

Abstract

We introduce the Segment Anything (SA) project: a new task, model, and dataset for image segmentation. Using our efficient model in a data collection loop, we built the largest segmentation dataset to date (by far), with over 1 billion masks on 11M licensed and privacy respecting images. The model is designed and trained to be promptable, so it can transfer zero-shot to new image distributions and tasks. We evaluate its capabilities on numerous tasks and find that its zero-shot performance is impressive -- often competitive with or even superior to prior fully supervised results. We are releasing the Segment Anything Model (SAM) and corresponding dataset (SA-1B) of 1B masks and 11M images at https://segment-anything.com to foster research into foundation models for computer vision.