Extract
Self-Deployed Container
Release Notes

Release Notes

v3.7.0

  • More accurate OCR on scanned documents. You may see higher peak memory while processing scanned documents. Check your allocation against your own documents before upgrading
  • Four new request options, all defaulting to v3.6.0 behavior: ocr, language, pages, and return_absolute_pdf_page_numbers
  • New optional MAX_OCR_PAGE_MEGAPIXELS setting, which rejects a document containing a page above a size limit you choose
  • Figure extraction tables use the same column names for every chart type, and a bar chart with a line over it and two value axes now returns two tables. Update any code that reads figure tables by column name or expects one table per figure

v3.6.0

  • Structured error responses with machine-readable error codes for document processing failures
  • Encrypted PDF detection

Bug fixes.