# brendenlake/omniglot

Omniglot data set for one-shot learning

Repository: https://github.com/brendenlake/omniglot
Canonical: https://ross.abutalabs.com/products/omniglot
Language: MATLAB
License: MIT
License Family: permissive
Last push: 2023-02-01T21:28:32+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3971, "days_push": 1309, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1421, forks 356 (observed 2026-08-28T04:04:40.675999+00:00)

## What it is
The Omniglot dataset for one-shot learning, containing 1,623 handwritten characters from 50 alphabets, each drawn by 20 people with paired stroke coordinate data. It includes background/evaluation splits and Python starter code for benchmarking human-like learning algorithms.

## Use cases
- benchmark one-shot learning algorithms
- train few-shot image classification models
- research meta-learning with minimal training data
- model human-like concept learning from few examples
- work with handwritten character stroke data
- evaluate generative models of handwriting

## When to choose
- you need a standard benchmark for one-shot or few-shot learning
- you want both raster images and stroke sequences for characters
- you are reproducing results from the Omniglot challenge papers

## When to avoid
- you need large-scale training data with many examples per class
- you need production-ready code rather than a research dataset
- you need natural images rather than handwritten characters

## Facets
- artifact type: dataset
- maturity: stable
- function: machine-learning, data-science, computer-vision
- domain: machine-learning, deep-learning, computer-vision
- platform: python, cross-platform
- tags: one-shot-learning, handwritten-characters, benchmark-dataset, few-shot-learning, stroke-data, meta-learning, natural-language-processing

## Member repositories
- brendenlake/omniglot (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:40.675999+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:37:44.392720+00:00, confidence not recorded.
  - readme: https://github.com/brendenlake/omniglot (fetched 2026-08-28T04:04:40.675999+00:00, sha 23928a31c114)
- Data as of 2026-08-30T08:39:29.467469+00:00.