document-ocr

What problem does it solve?

Build an end-to-end OCR ingestion pipeline that converts scanned documents and images into structured, searchable text with per-token confidence and coordinates, simplifying data capture from forms and handwritten notes.

Core Features & Use Cases

• OCR-driven ingestion: Import PDFs, images, and multi-page documents; normalize scans; detect and segment form regions.
• Structured output: Produce text with bounding boxes and per-region mappings (e.g., headers, blocks, lines, checkboxes) and embed confidence data.
• Use Case: Use this when digitizing handwritten forms or notepads to extract sections like Observations, Health/Emotional Status, Review of Services, Goals, Additional Notes, Follow-up Tasks, and Signature blocks for archival and analytics.

Quick Start

Run the pipeline on a sample document to generate ocr_output.json with the attached data structure, along with preprocess_pipeline.md and test_corpus_guidelines.md.