name: image-ocr description: "Extract text from images using Tesseract OCR (supports Chinese, English, and other languages)." version: 1.1.0 author: Hermes Agent license: MIT platforms: [windows, macos, linux] metadata: hermes: tags: [OCR, Image, Text-Extraction, Chinese, English, Cross-Platform]

Image OCR (Tesseract)

Extract text from images using Tesseract OCR engine with Python wrapper. Cross-platform support for Windows, macOS, and Linux.

Prerequisites

• Tesseract OCR (v5.0+)
• Python packages: pytesseract, Pillow
• Language packs: Chinese (chi_sim) and English (eng) recommended

Install Tesseract

Install Python Dependencies

pip install pytesseract Pillow

Usage

Basic OCR

python scripts/ocr.py <image_path> [-l LANGUAGE] [--psm PSM]

Examples:

# Auto-detect Tesseract, use default language (chi_sim+eng)
python scripts/ocr.py screenshot.png

# Specify language
python scripts/ocr.py document.jpg -l chi_sim+eng

# Specify page segmentation mode
python scripts/ocr.py photo.jpg -l eng --psm 6

# Specify Tesseract path manually
python scripts/ocr.py image.png --tesseract-path "C:\Program Files\Tesseract-OCR"

# List installed language packs
python scripts/ocr.py --list-langs

Tesseract Discovery (auto-detection order)

The script finds Tesseract automatically in this order:

1. --tesseract-path CLI argument
2. TESSERACT_PATH environment variable
3. Config file (~/.config/image-ocr/config.json or %APPDATA%/image-ocr/config.json)
4. Common install locations:
5. Windows: C:\Program Files\Tesseract-OCR, C:\Program Files (x86)\Tesseract-OCR
6. macOS: /usr/local/bin/tesseract, /opt/homebrew/bin/tesseract
7. Linux: /usr/bin/tesseract, /usr/local/bin/tesseract
8. System PATH

Persistent Configuration

Create a JSON config file to set a persistent Tesseract path:

Windows: %APPDATA%\image-ocr\config.json macOS/Linux: ~/.config/image-ocr/config.json

{
  "tesseract_path": "/path/to/tesseract"
}

Supported Languages

Combine multiple languages: chi_sim+eng

Script Location

scripts/ocr.py - Main OCR script
小葱技能7w4.net有完整的技能分类。

Notes

• For best results, ensure images are clear and well-lit
• Chinese recognition requires chi_sim language pack
• Output is plain text; use --psm flag for specific page segmentation modes
• Verbose diagnostic info (Tesseract version, path, language) is printed to stderr

这是一款实用的图像文字识别工具，中文识别效果好，支持多平台使用。安装和配置说明详细，错误提示友好。但文档与实际脚本名称存在不一致，版本信息也有出入。对于普通用户而言功能是可靠的，但使用前需注意阅读最新文档确认正确的脚本名称。