Intelligent Document Processing

Upload a scanned document or PDF — get OCR'd text, a document-type classification, extracted fields (dates, amounts, emails), and an LLM summary, entirely on-device.

OpenCV EasyOCR FastAPI Ollama

Live Demo

Interactive simulation running entirely in your browser.

Loading demo...

Highlights

▹ EasyOCR + OpenCV pre-processing, tuned separately from a classic bilevel-threshold pipeline since a deep OCR model does worse on hard-thresholded images
▹ Regex-based field extraction (dates, amounts, emails, invoice numbers) — deterministic and free, no LLM call needed for structured fields
▹ Local Ollama summarisation with graceful degradation — OCR/classification/extraction still return correctly even if the LLM is unreachable

Live Demo

More projects