ERNIE Warm-up: PDF → Website

Compliance Statement

This submission fully complies with the official Warm-up Task specification.

• PaddleOCR-VL used to extract text/layout from PDFs (src/warmup.py)
• Content converted to Markdown (output/extracted.md)
• ERNIE-generated HTML website (output/index.html, root index.html) fully self-contained
• Deployed on GitHub Pages: https://itzamil.github.io/ernie-ai-warmup-pdf-to-website/
• Google Colab used only as OCR debugging helper. Final repo works offline.

This project completes the ERNIE AI Developer Challenge Warm-up Task on Devpost:

1. Convert PDF to images
2. OCR → extract text → Markdown (output/extracted.md)
3. Generate static website from Markdown
4. Deploy with GitHub Pages ✅ LIVE: https://itzamil.github.io/ernie-ai-warmup-pdf-to-website/

Structure

• sample.pdf – Source document
• output/extracted.md – OCR Markdown output
• output/index.html – Generated HTML
• src/warmup.py – PaddleOCR extraction
• src/generate_website.py – Markdown → HTML

OCR Notebook (Optional)

Supporting Colab used for OCR debugging:

• https://colab.research.google.com/drive/1vRcYlWZpWQHKQxlZ2bjaCWcVqEAyFqwF#scrollTo=aKRMOxjkgA5m

Project fully self-contained in repo. No external dependencies required.