# mindee/doctr

docTR (Document Text Recognition) - a seamless, high-performing & accessible library for OCR-related tasks powered by Deep Learning. Ongoing development and maintenance by t2k.

Repository: https://github.com/mindee/doctr
Canonical: https://ross.abutalabs.com/products/doctr
Homepage: https://mindee.github.io/doctr/
Language: Python
License: Apache-2.0
License Family: permissive
Topics: ocr, deep-learning, document-recognition, tensorflow2, text-detection-recognition, text-detection, text-recognition, optical-character-recognition, pytorch
Last push: 2026-08-21T14:16:15+00:00

## Health v2 (maintenance only)
Score: 90/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 98, release rhythm 74, longevity 100
- inputs: {"age_days": 2063, "days_push": 12, "days_rel": 12, "gap_med": 140, "n_releases_24m": 6}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 6315, forks 673 (observed 2026-08-28T04:09:41.710923+00:00)

## What it is
docTR is a Python OCR library that extracts text from documents and images using a two-stage deep learning approach: text detection followed by text recognition. It offers pretrained models in PyTorch and TensorFlow, layout analysis, and a simple 3-line API for end-to-end text extraction.

## Use cases
- extract text from scanned documents
- run ocr on pdfs and images in python
- localize and recognize words in document images
- detect tables and layout regions in documents
- benchmark my own text recognition model against state-of-the-art ocr
- deploy an ocr api service
- convert receipts or invoices to text

## When to choose
- you need high-accuracy OCR comparable to cloud services like AWS Textract but self-hosted
- you want a choice of detection and recognition architectures with pretrained weights
- you need both CPU and GPU inference with a lightweight dependency footprint
- you want layout analysis (tables, figures, headers) alongside text extraction

## When to avoid
- you need handwriting transcription or full-page document understanding beyond text/layout
- you want a turnkey GUI application rather than a Python library
- you need OCR for non-Latin scripts not covered by the model zoo
- you prefer a cloud OCR API with zero model management

## Facets
- artifact type: library
- maturity: active
- function: ocr, computer-vision, deep-learning, machine-learning, image-processing
- domain: computer-vision, image-processing, deep-learning, machine-learning
- platform: python, cross-platform
- tags: text-detection, text-recognition, pytorch, tensorflow, document-parsing, layout-analysis, pretrained-models, natural-language-processing, gpu, docker

## Member repositories
- mindee/doctr (main) score 90

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:41.710923+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 2026-08-29T17:46:12.294598+00:00, confidence not recorded.
  - readme: https://github.com/mindee/doctr (fetched 2026-08-28T04:09:41.710923+00:00, sha 2a1225f61590)
  - homepage: https://mindee.github.io/doctr/ (fetched 2026-08-29T08:42:37.541026+00:00, sha 7e3ffab1342c)
- Data as of 2026-08-30T08:39:29.467469+00:00.