Tesseract Versions Save

This is the third release candidate of Tesseract 5.0.0.

• Improve training messages
• Add RowAttributes getter to PageIterator

See also list of all changes.

This is a new stable release of Tesseract 4.1.

• Fix broken autoconf build (issue #3642)

See also list of all changes.

This is a new stable release of Tesseract 4.1.

Note: The autoconf build is broken (see issue #3642), so please use 4.1.3.

• Allow line images with larger width for training
• Bug fixes
• Build updates and fixes

See also list of all changes.

This is the second release candidate of Tesseract 5.0.0.

• Fix regression for OCR with more than one model file
• Bug fixes
• Optimizations

See also list of all changes.

This is the first release candidate of Tesseract 5.0.0.

• Enable fast float32 LSTM by default
• Switch to NFC normalisation everywhere
• Remove banner message
• Disable music staff detection and removal
• Add new command line option --loglevel
• Bug fixes

See also list of all changes.

This is a new pre-release of Tesseract 5.0.0.

• Bug fixes
• Extend URI support for Tesseract with libcurl
• Rename processed TIFF output file and add page number if needed

See also list of all changes.

This is a new pre-release of Tesseract 5.0.0.

• Bug fixes
• Modernize more code
• More options for binarization
• Improved support for ARM NEON
• No longer depends on Abseil for unit tests
• Support float for model training and text recognition (faster, requires less RAM)

See also list of all changes.

This is a new pre-release of Tesseract 5.0.0.

• Replaced all remaining STRING by std::string
• Replaced lots of GenericVector by std::vector
• Replaced all malloc / free by C++ code
• Modernized and formatted code

See also list of all changes.

This is a new pre-release of Tesseract 5.0.0.

It has massive changes in the public API which is a great step towards a final 5.0.0. All unit tests pass, but because of those changes more practical experience is needed.

• the public API no longer uses proprietary data types GenericVector, STRING
• pdf.ttf is no longer needed because it is now integrated into the code

See also list of all changes.

This is a new pre-release of Tesseract 5.0.0.

It is considered to be production ready for end users, but nevertheless not stable because more incompatible API changes are planned.

• improved performance (also on ARM / ARM64)
• improved unit tests
• many fixes
• faster flat build with automake
• support for latest macOS (including new M1 processor)

See also list of all changes.