Home·Curated Catalog·Computer Vision
👁 Curated Catalog · Computer Vision

SA-1B — Segment Anything

11M licensed images with 1.1 billion segmentation masks from Meta AI.

LQS 89 · gold ⚠ Research-only 1.1B segmentation masks 11 TB JPG · JSON Released 2023
Browse commercial Computer Vision → Visit original source ↗
Source: ai.meta.com · maintained by Meta AI Research
1.1B
segmentation masks
11 TB
Size on disk
89
LQS · gold
2023
First released

About this dataset

SA-1B is the dataset released alongside Meta AI's Segment Anything Model (SAM). It contains 11M high-resolution licensed photos paired with 1.1 billion high-quality automatically-generated segmentation masks — approximately 100 masks per image. The masks cover objects at multiple scales and enable universal segmentation models.

Maintainer
Meta AI Research
License
SA-1B Research License
Formats
JPG · JSON
Paper
Read on arxiv.org →

LabelSets Quality Score

LQS is our 7-dimension quality score, computed from the dataset's published statistics. See methodology →

89
out of 100
gold tier

High-quality dataset across most dimensions

Composite score computed from the 7 dimensions below: completeness, uniqueness, validation health, size adequacy, format compliance, label density, and class balance.

Completeness 85
No public completeness metric; using prior for 'automated' datasets.
Uniqueness 93
Exact-hash deduplication documented by maintainer.
Validation 75
Labels generated by a trained model (e.g., automatic mask generation).
Size adequacy 100
1,100,000,000 segments — exceeds 100,000 adequacy target for Computer Vision.
Format compliance 95
Industry-standard format — drop-in compatible with mainstream tooling.
Label density 100
Average 100.0 labels per item (high density).
Class balance 90
Near-uniform class distribution.

What it's used for

Common tasks and benchmarks where SA-1B — Segment Anything is the default or competitive choice.

Sample statistics

What's actually in the dataset — from the maintainer's published stats.

11M images, avg 100 masks/image, 1.1B total masks. Images at ~1500×2250 resolution. Masks mostly high-quality (IoU filtered).

License

SA-1B — Segment Anything is distributed under SA-1B Research License. This is a third-party public dataset; LabelSets indexes and scores it but does not host or redistribute the data. Always verify current license terms with the maintainer before commercial use.

Heads up: this dataset's license restricts commercial use. If you need computer vision data for production, check LabelSets' paid datasets below — every listing has an explicit commercial license.

Need commercial-licensed Computer Vision data?

LabelSets sellers offer paid computer vision datasets with what public datasets often can't give you:

Browse paid Computer Vision → Sell your dataset

Similar public datasets

Other entries in the Computer Vision catalog.

COCO — Common Objects in Context
330K images · CC BY 4.0
ImageNet
14.2M images · Custom (non-commercial research)
Open Images V7
9M images · CC BY 4.0 (annotations) / varies (images)
CIFAR-100
60K images · MIT-style (unrestricted research use)

Frequently Asked Questions

SA-1B — Segment Anything is distributed under SA-1B Research License, which restricts commercial use. For a commercially-licensed alternative in computer vision, see LabelSets' paid datasets.
SA-1B — Segment Anything contains 1,100,000,000 segmentation masks. 11M images, avg 100 masks/image, 1.1B total masks. Images at ~1500×2250 resolution. Masks mostly high-quality (IoU filtered).
SA-1B — Segment Anything is maintained by Meta AI Research and is available at https://ai.meta.com/datasets/segment-anything-downloads/. LabelSets indexes and scores this dataset for discoverability but does not redistribute it.
LQS is a 7-dimension quality score (completeness, uniqueness, validation, size adequacy, format compliance, label density, class balance) computed from the dataset's published statistics. Composite scores map to tiers: platinum (≥90), gold (≥75), silver (≥60), bronze (<60). Read the full methodology.