Що нового

JSON Schema extraction — define a schema and get structured data from any document

Citations — bounding box coordinates for each extracted field

HEIC image support

rekognita-fast speed increased by 30%

Improved table recognition accuracy by 12%

Graph structure — full element graph of document

Word/PDF generation — restructured documents as output

Async Python SDK client

Fixed nested table recognition

Fixed encoding for non-Latin characters

Webhooks with HMAC-SHA256 signing

Multi-page TIFF support

page_range parameter for partial processing

rekognita-accurate quality +8% for financial documents

API response time -20%

Three processing models: rekognita-fast, rekognita-balanced, rekognita-accurate

25+ language OCR support

HTML output format

Batch processing up to 50 docs per request

Fixed PDF multi-page processing

Initial public release

REST API with authentication

Document OCR and text extraction

JSON and plain text output

Node.js and Python SDK