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

ADE20K

Scene parsing benchmark with 25K images and pixel-level masks across 3,500+ object classes.

LQS 86 · gold ✓ Commercial OK 25.6K images 3.8 GB JPG · PNG Released 2017
Browse commercial Computer Vision → Visit original source ↗
Source: groups.csail.mit.edu · maintained by MIT CSAIL
25.6K
images
3.8 GB
Size on disk
86
LQS · gold
2017
First released

About this dataset

ADE20K is MIT CSAIL's scene parsing benchmark. 25,574 images with dense pixel-level segmentation masks covering 3,500+ object classes and parts — objects like 'ceiling lamp' or 'microwave door' — along with 150 stuff/thing categories in the standard evaluation subset.

Maintainer
License
Formats
JPG · PNG

LabelSets Quality Score

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

86
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 95
No public completeness metric; using prior for 'expert_curated' datasets.
Uniqueness 93
Exact-hash deduplication documented by maintainer.
Validation 92
Labels produced by domain experts or trained annotators.
Size adequacy 68
25,574 images — below 100,000 target for Computer Vision, but usable.
Format compliance 95
Industry-standard format — drop-in compatible with mainstream tooling.
Label density 93
Average 20.0 labels per item (high density).
Class balance 58
Long-tail distribution — dominant classes overrepresented.

What it's used for

Common tasks and benchmarks where ADE20K is the default or competitive choice.

Sample statistics

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

25,574 images, 3,500+ fine object/part classes, 150 semantic categories in the benchmark subset. Pixel-level masks.

License

ADE20K is distributed under BSD-3-Clause. 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.

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.

Frequently Asked Questions

ADE20K is distributed under BSD-3-Clause, which generally permits commercial use. Always verify the current license terms with the maintainer (MIT CSAIL) before using in a commercial product.
ADE20K contains 25,574 images. 25,574 images, 3,500+ fine object/part classes, 150 semantic categories in the benchmark subset. Pixel-level masks.
ADE20K is maintained by MIT CSAIL and is available at https://groups.csail.mit.edu/vision/datasets/ADE20K/. 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.