Automated Document Information Extraction (Key-Value) via Donut Model
OCR-free visual document transformer reading crumpled receipts, invoices, and utility bills directly into structured JSON.
Project Overview
Utilizes the cutting-edge Donut (Document Understanding Transformer) neural architecture that bypasses brittle legacy OCR engines. Takes raw document photos, decodes typography and visual hierarchy directly via an encoder-decoder transformer, and outputs clean structured JSON.
Utilizes the cutting-edge Donut (Document Understanding Transformer) neural architecture that bypasses brittle legacy OCR engines. Takes raw document photos, decodes typography and visual hierarchy directly via an encoder-decoder transformer, and outputs clean structured JSON.
OCR-free visual document transformer reading crumpled receipts, invoices, and utility bills directly into structured JSON.