📄 DeepSeek-OCR PDF Parser by Jatevo LLM Inference

Upload a PDF to extract text and convert to Markdown using DeepSeek-OCR.
Each page is processed sequentially and combined into a single markdown document.

✨ Features

• 🖼️ Image Embedding - Charts, graphs, and figures embedded directly in markdown
• 📝 Text Extraction - All text content from images and charts extracted
• 📊 Table Support - Tables converted to markdown format
• 🔍 Object Detection - Locate specific elements in documents
• 🎯 Multiple Models - Choose speed vs. accuracy trade-off

📏 Model Sizes

• Tiny — Fastest, lower accuracy (512×512) - Best for large PDFs (30+ pages)
• Small — Fast, good accuracy (640×640) - Good for 15-30 pages
• Base — Balanced performance (1024×1024) - Good for 10-20 pages
• Large — Best accuracy, slower (1280×1280) - Best for <10 pages
• Gundam (Recommended) — Optimized for documents (1024 base, 640 image, crop mode)

💡 Tips

• Enable "Embed Images" to include charts/figures (recommended)
• Use Tiny or Small model for large PDFs (20+ pages)
• Processing time: ~2-5 seconds per page depending on model
• Maximum recommended: 50 pages at once
• Image embedding increases file size (~1-2MB per page with images)