# diffgram/diffgram

The AI Datastore for Schemas, BLOBs, and Predictions. Use with your apps or integrate built-in Human Supervision, Data Workflow, and UI Catalog to get the most value out of your AI Data.

Repository: https://github.com/diffgram/diffgram
Canonical: https://ross.abutalabs.com/products/diffgram
Homepage: https://diffgram.com
Language: Python
License: NOASSERTION
License Family: other
Topics: annotation, annotation-tool, training-data, video-annotation, data-annotation, kubernetes, data-science, data-analytics, image-annotation, machine-learning, deep-learning, data, annotations, datasets, labeling, datastore
Last push: 2026-06-22T07:33:28+00:00

## Health v2 (maintenance only)
Score: 62/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 88, release rhythm 8, longevity 100
- inputs: {"age_days": 2918, "days_push": 72, "days_rel": 689, "gap_med": null, "n_releases_24m": 1}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1909, forks 135 (observed 2026-08-28T04:05:53.030194+00:00)

## What it is
Diffgram is a self-hosted AI datastore for managing schemas, BLOBs, and predictions, with built-in human supervision (data labeling), data workflow, and a UI catalog for exploring AI data. It supports annotating many media types including image, video, text, audio, 3D, geospatial, and conversational/LLM data.

## Use cases
- label images and videos for training data
- annotate text and audio datasets for machine learning
- manage training data for AI apps in one place
- store and explore AI predictions with a UI catalog
- scale annotation teams with a data labeling workflow
- handle PII-compliant AI data self-hosted
- annotate 3D and geospatial data
- review and correct LLM conversational outputs

## When to choose
- you need a self-hosted labeling platform across many media types
- you want an integrated datastore plus annotation workflow rather than a standalone tool
- your team needs to scale human supervision with UI-based data exploration
- you need control over sensitive or PII training data

## When to avoid
- you only need a lightweight single-purpose image annotator
- you want a fully permissive open-source license (Diffgram uses its own DLv2 license)
- you need fully automated labeling without human-in-the-loop
- you prefer a managed SaaS without self-hosting overhead

## Facets
- artifact type: application
- maturity: active
- function: machine-learning, data-science, image-processing, video-processing, nlp, audio-processing, geospatial, self-hosted
- domain: machine-learning, data-science, computer-vision, deep-learning, artificial-intelligence
- platform: python, self-hosted, cross-platform
- tags: data-labeling, annotation-tool, training-data, data-annotation, human-in-the-loop, ai-datastore, computer-vision-annotation, video-annotation, text-annotation, audio-annotation, 3d-annotation, geospatial-annotation, llm-annotation, dataset-management, commercial-open-source, natural-language-processing, docker, kubernetes, web-server

## Member repositories
- diffgram/diffgram (main) score 62

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:53.030194+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 2026-08-30T03:11:06.174277+00:00, confidence not recorded.
  - readme: https://github.com/diffgram/diffgram (fetched 2026-08-28T04:05:53.030194+00:00, sha 92a3e2b71bb9)
  - homepage: https://diffgram.com (fetched 2026-08-29T10:50:06.190082+00:00, sha 48856ee6523e)
- Data as of 2026-08-30T08:39:29.467469+00:00.