Skip to content
TakoAPI
Back to skills

openocr-skill

by topdu

Image & Video Generation
1.2kDownloads reported by clawskills.sh
0Views
0 likesSign in to like this skill
—Not yet rated
Your rating

Brief

Extract text from images, documents and scanned PDFs using OpenOCR.

Description

What This Skill Does

OpenOCR extracts text from images, scanned documents, and PDFs using a lightweight OCR engine. It supports text detection, recognition, formula and table recognition via a 0.1B parameter VLM, and full document parsing with layout analysis. Runs on CPU by default with optional GPU acceleration.

Runs entirely locally with a 0.1B parameter model on CPU, avoiding API costs and data privacy concerns of cloud OCR services.

When to Use It

  • Extracting text from scanned invoices or receipts
  • Converting scanned academic papers to searchable Markdown
  • Recognizing mathematical formulas from textbook photos
  • Pulling structured data out of table screenshots
  • Batch processing a folder of document images to plain text

Example Workflow

INPUT

User asks: Parse this scanned PDF and convert it to Markdown

AGENT
  1. 1

    Initialize OpenOCR with the doc task and layout detection enabled

  2. 2

    Load the PDF and convert pages to images internally

  3. 3

    Run layout analysis to identify text blocks, formulas, and tables

  4. 4

    Recognize content in each block using the UniRec VLM

  5. 5

    Save the structured output as a Markdown file per page

OUTPUT

Markdown files with preserved document structure including headings, paragraphs, recognized formulas, and tables

Installation

clawhub install topdu/openocr-skill

Run the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.

Links