deepseek_ocr_app

Why we included this project

This React + FastAPI app wraps the DeepSeek-OCR model in a ready-made web interface, so you can point it at a scanned page or a multi-page PDF and get editable text back without scripting the model yourself. It handles single-image OCR and full PDF processing, with four working modes that range from plain text extraction to finding specific terms and drawing bounding boxes around them. The export side is where it earns its keep: results come out as Markdown, HTML, DOCX, or JSON, and tables, LaTeX formulas, and embedded images survive the trip, which matters when you're digitizing academic papers or converting business reports. For a team that wants a self-hosted OCR endpoint with a clean UI, it's a fast way to stand up document digitization or data extraction without building the frontend and API plumbing from scratch.