OCR Vietnamese

😫 The Pain Point

You have scanned contracts or photos of documents in Vietnamese. You need the text searchable and editable. Retyping manually is slow and error-prone.

🚀 Agentic Solution

An OCR Tool optimized for Vietnamese text with high accuracy.

Key Features:

• Vietnamese Language Pack: Trained for VN characters (ă, â, đ, ơ, ư).
• Image Preprocessing: Enhance contrast for better recognition.
• Batch Processing: Extract text from multiple images.

⚔️ Phase 1: Commander (Quick Fix)

For quick OCR.

Prompt:

“I have a folder scans with images of Vietnamese documents. Write a Python script using pytesseract to:

1. Preprocess: Convert to grayscale, increase contrast.
2. OCR: Extract text using Vietnamese language pack (vie).
3. Output: Save text to {filename}.txt for each image.

Print progress. Handle unreadable images (skip with warning).”

Result: Editable text from all scanned documents.

🏗️ Phase 2: Architect (Permanent Tool)

Engineering Prompt:

**Role:** Python Tool Developer
**Task:** Create a "Vietnamese OCR Tool".

**Requirements:**
1.  **GUI:**
    *   Select image or folder.
    *   Language dropdown (vie, eng, vie+eng).
    *   Preprocessing options (contrast, rotate).
    *   Preview extracted text.
    *   Export as TXT or DOCX.

2.  **Logic:**
    *   Use pytesseract with tessdata.
    *   Image preprocessing with Pillow.
    *   Confidence scoring.

3.  **Deliverables:**
    *   `ocr_vietnamese.py`
    *   `run.bat`, `run.sh`
    *   `requirements.txt`

🧠 Prompt Decoding

• Tesseract vie: Must download Vietnamese language data separately.

🛠️ Instructions

1. Install Tesseract OCR engine.
2. Download Vietnamese language pack.
3. Install: pip install pytesseract pillow
4. Copy Prompt → Run.