# kotaro-kinoshita/yomitoku

YomiTokuはAIを活用した日本語文書解析エンジンを提供するPythonパッケージです。 Yomitoku is an AI-powered document image analysis package designed specifically for the Japanese language.

Repository: https://github.com/kotaro-kinoshita/yomitoku
Canonical: https://ross.abutalabs.com/products/yomitoku
Homepage: https://kotaro-kinoshita.github.io/yomitoku/
Language: Python
License Family: other
Topics: deep-learning, layout-analysis, ocr, python, pytorch
Last push: 2026-08-18T04:16:13+00:00

## Health v2 (maintenance only)
Score: 87/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 98, release rhythm 95, longevity 48
- inputs: {"age_days": 672, "days_push": 15, "days_rel": 33, "gap_med": 15.0, "n_releases_24m": 31}
- 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 1579, forks 58 (observed 2026-08-28T04:05:06.736453+00:00)

## What it is
YomiToku is an AI-powered document image analysis engine specialized for Japanese, providing full-text OCR, layout analysis, table structure recognition, and reading-order estimation. It converts document images into HTML, Markdown, JSON, CSV, or searchable PDFs, running on GPU or CPU with PyTorch-based models.

## Use cases
- extract text from Japanese document images with ocr
- convert scanned pdfs to searchable pdf or markdown
- analyze document layout and reading order
- recognize table structure in document images
- read handwritten or vertical Japanese text
- extract figures and tables from document images

## When to choose
- you need high-accuracy OCR for Japanese documents including handwriting and vertical layouts
- you want structured output (markdown, html, json, csv) preserving document semantics
- you need table structure recognition and layout analysis, not just raw text
- you want local inference without sending images to external servers

## When to avoid
- you need a permissive open-source license for commercial use - commercial use requires a paid license
- your documents are primarily non-Japanese and non-English
- you need a lightweight pure-CPU pipeline without model downloads
- you only need simple OCR without layout or table analysis

## Facets
- artifact type: library
- maturity: active
- function: ocr, image-processing, machine-learning, deep-learning, pdf, parser
- domain: computer-vision, pdf, machine-learning, developer-tools
- platform: python, windows, cli
- tags: document-ai, layout-analysis, japanese, table-structure-recognition, searchable-pdf, pytorch, handwriting-recognition, natural-language-processing, linux, macos, gpu

## Member repositories
- kotaro-kinoshita/yomitoku (main) score 87

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:06.736453+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 2026-08-30T03:56:43.765391+00:00, confidence not recorded.
  - readme: https://github.com/kotaro-kinoshita/yomitoku (fetched 2026-08-28T04:05:06.736453+00:00, sha 3f04c0fd1401)
  - homepage: https://kotaro-kinoshita.github.io/yomitoku/ (fetched 2026-08-29T11:27:09.505482+00:00, sha a335c36b03bc)
- Data as of 2026-08-30T08:39:29.467469+00:00.