Fix:

• Page Segmentation Mode works now.

Thanks to @TheBlune for PR.

New Feature:

• Support Blob format in browser.

Thanks for contributions from @nisarhassan12, @iscoolapp, @tmcw and @monkeywithacupcake !!

Breaking Changes:

• Names of parameters are adjusted to make it more consistent.

Major update:

• Now you can recognize only part of the image, check example HERE
• detect() and recognize() is now thenable, check HERE

New features:

• Add Open Collective
• Add Gitpod support

Thanks to @HoldYourWaffle, @tmcw and @antonrifco for PRs.
Major changes:

• Default OCR Engine Mode changes to LSTM_ONLY

Minor changes:

• Remove LFS support
• Enable new image formats
  • Buffer (Node.js only)
  • Base64 string
• Fix process issue (for angular.js)

• Update tesseract.js-utils to support loading language data as a binary array.

• Dump hocr, tsv, box, unlv, osd information in final result

• Ignore DOMException in loadLang for now (a weird issue in Google Chrome)
• Update documents about how to train your own data.
• Minor restructure tests folder
• Fix lint error
• Add babel-loader to transpile code to pure es5 (for browser compatibility)

Add pdf_auto_download and pdf_bin parameters to support custom handling of pdf binary data.

• Fix ocr engine mode issue
• Fix page seg mode issue
• Support pdf output

• Remove crossorigin.me in loadImage to prevent security issues.
• Add application/javascript type when initializing Worker
• Enhance README