docTR — The Complete Guide: Deep-Learning OCR for Documents and Forms
docTR, from Mindee, is a Python OCR library that works in two stages: first it detects the text regions, then it recognises them. The licence is Apache 2.0, processing is local, and no registration is needed.
docTR
Mindee
Open-source deep-learning OCR library: text detection and recognition in PDFs and images, including rotated pages and table region detection.
Get it from the vendorWhat you get
- Varied input — PDFs, images and even web pages.
- Rotated pages — correct recognition even when the scan is skewed.
- Region types — title, text, table and page header.
- Structured JSON output — including the position of every word.
- Deployment — a demo app, a FastAPI template and Docker images with GPU support.
Who it suits
docTR is aimed at engineers and developers building a recurring pipeline: extracting data from forms, material test certificates or scanned data sheets, and feeding it into a database. There is no graphical interface, and it needs Python 3.11 or later and PyTorch. The record does not confirm Hebrew support, so test before using it on Hebrew documents.
Related tools
For layout analysis and cell-level tables — PaddleOCR. For converting documents to Markdown — Docling. For a simpler library for single images — EasyOCR.
PaddleOCR
PaddlePaddle (Baidu)
Open-source Python OCR and document parsing toolkit: text recognition in dozens of languages, page layout, tables and PDF to Markdown or JSON.
Get it from the vendor
Docling
Docling Project (originally IBM Research)
Open-source tool that converts PDF, DOCX and scanned images to Markdown or JSON, with table structure recognition and OCR, running locally.
Get it from the vendor
EasyOCR
Jaided AI
Simple Python library for recognising text in images in more than 80 languages, with each word’s position and confidence — in a few lines of code.
Get it from the vendor
Python
Python Software Foundation
The language behind most modern engineering computation, and its full ecosystem.
Get it from the vendorThe bottom line
A solid foundation for an in-house OCR service that processes documents on a regular basis. For a one-off job with a graphical interface, choose a desktop tool.