APRIL 5, 2023

SA-1B Dataset

Segment Anything 1 Billion (SA-1B) is a dataset designed for training general-purpose object segmentation models from open world images. The dataset was introduced in our paper “Segment Anything”.


SA-1B consists of 11M diverse, high-resolution, privacy protecting images and 1.1B high-quality segmentation masks that were collected with our data engine. It is intended to be used for computer vision research for the purposes permitted under our Data License.

The images are licensed from a large photo company. The 1.1B masks were produced using our data engine, all of which were generated fully automatically by the Segment Anything Model (SAM). Please refer to the paper for more details on the mask generation process.

SA-1B V1.0

Key Application

Computer Vision, Segmentation

Intended Use Cases
  • Research purposes only

  • Train and evaluate generic object segmentation models

  • Allow access to a privacy protecting and copyright friendly large-scale image dataset

Primary Data Type

Images, Mask annotations

Data Function

Training, testing

Dataset Characteristics

Total number of images: 11M

Total number of masks: 1.1B

Average masks per image: 100

Average image resolution: 1500×2250 pixels

NOTE: There are no class labels for the images or mask annotations.


Class agnostic mask annotations

Nature Of Content

The underlying images are licensed from a large photo company. The images vary in subject matter. Common themes of the images include: locations, objects, scenes. Masks range from large scale objects such as buildings to fine grained details such as door handles.

Privacy PII

Faces and license plates de-identified


Limited; see full license language for use

Access Cost

Open access

Data Collection

Data sources

Images licensed from a photo company.

Masks generated by the Segment Anything Model (SAM).

Data selection

Images were selected based on their content. The images are photos taken from a camera, i.e. not artwork.

Sampling Methods


Geographic distribution

Labeling Methods

Automatically generated masks (more details in the Segment Anything paper)

Label types

Masks are provided in the COCO run-length encoding (RLE) annotation format

Labeling procedure - Automatic

The final mask annotations we are releasing were generated automatically. To train the model used for automatic annotation, we first collected mask annotations from expert human annotators using an interactive model in the loop process. Please refer to our paper for more details.

Validation Methods

A random sample of mask annotations were reviewed and validated by human annotators.

Please email segment-anything@meta.com or report any issues via the feedback form on our website segment-anything.com