skills.rest
damionrashford
damionrashfordCOMMUNITY
Explore all skills by this author →
GitHub

media-ocr-ai

Extract text and structure from images and PDFs using multiple OCR backends.

15|4|Updated Apr 18, 2026
Search Tags:#ocr#multimodel#tesseract#handwriting#paddleocr#easyocr#trocr
One-click install
npx skills add https://github.com/damionrashford/media-os --skill media-ocr-ai
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: media-ocr-ai
Source: https://github.com/damionrashford/media-os/tree/main/skills/media-ocr-ai
Command: npx skills add https://github.com/damionrashford/media-os --skill media-ocr-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires paddlepaddle>=2.6, paddleocr>=2.7, easyocr>=1.7, pytesseract>=0.3.10, transformers>=4.40, torch>=2.2, opencv-python>=4.9, numpy>=1.24, pillow>=10.0, and includes scripts (resource) and references (resource) components.

What problem does it solve?

OCR can be slow and error-prone when handling diverse documents, languages, and formats. This skill provides a unified, offline OCR workflow that lets you choose among multiple open-source backends (PaddleOCR, EasyOCR, Tesseract) and the handwriting-focused TrOCR to extract text and structure from images and PDFs with consistent output.

Core Features & Use Cases

  • Multimodel backends: select PaddleOCR for structured layouts, EasyOCR for quick reads, Tesseract for broad language coverage, and TrOCR for handwriting.
  • Structured layout and table extraction: identify headers, paragraphs, and tables to produce usable JSON or CSV outputs.
  • Handwriting transcription: transform handwritten notes into editable text with line-level accuracy.
  • Multilingual support: handle documents containing multiple languages and scripts in one workflow.
  • Output flexibility: produce plain text, JSON blocks, TSV, or CSV for downstream pipelines.

Quick Start

Install the required backends with the install command and then run the extract or layout commands to process your documents.

Frequently Asked Questions about media-ocr-ai

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I extract text from images and PDFs with multiple OCR engines?▼

To extract text from images and PDFs, this skill uses PaddleOCR, EasyOCR, Tesseract, and TrOCR to process multilingual documents, tables, and handwriting, outputting plain text, JSON, TSV, or CSV formats.

What is the best way to transcribe handwriting from scanned documents?▼

Transcribing handwriting from scanned documents is handled by the TrOCR backend, which provides line-level accuracy to transform handwritten notes into editable text within a unified offline workflow.

Does this OCR workflow support structured layout and table extraction?▼

Yes, layout and table extraction is supported primarily through PaddleOCR, identifying headers, paragraphs, and tables to produce structured and usable JSON or CSV outputs for downstream pipelines.

Can I use Tesseract and EasyOCR for multilingual document processing?▼

Yes, you can use Tesseract for broad language coverage and EasyOCR for quick reads, allowing the workflow to handle documents containing multiple languages and scripts in one process.

Do I need to install PyTorch and PaddlePaddle to run this OCR skill?▼

Yes, you need PyTorch and PaddlePaddle installed, as the skill depends on these frameworks alongside transformers, OpenCV, and NumPy to run the diverse OCR backends and process images.

Related Skills

Explore More Data & Analytics Skills→
lotosbinlotosbin
14

screen-capture

Automate screen captures and OCR analysis across macOS and Python.

Community
Intermediate
omosb1-sysomosb1-sys

precision-extractor

Convert unstructured text into structured JSON/TOON assets with entity resolution.

Community
Advanced
AIDotNetAIDotNet
602

image-ocr

Extract text from images and scanned documents into plain text.

Official
Intermediate
LuYanFCPLuYanFCP

use-ocr

Extract text, tables, formulas, and handwriting from images and PDFs using local OCR tools.

Community
Intermediate