{"id":120548,"date":"2024-11-11T07:09:08","date_gmt":"2024-11-11T05:09:08","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/reliz-sistemy-raspoznavaniya-teksta-tesseract-5-5-0"},"modified":"2024-11-11T07:09:08","modified_gmt":"2024-11-11T05:09:08","slug":"reliz-sistemy-raspoznavaniya-teksta-tesseract-5-5-0","status":"publish","type":"post","link":"https:\/\/prohoster.info\/en\/blog\/news\/reliz-sistemy-raspoznavaniya-teksta-tesseract-5-5-0","title":{"rendered":"Release of the text recognition system Tesseract 5.5.0","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
The release of the optical character recognition system Tesseract 5.5.0 has been published, supporting Unicode and text recognition in more than 100 languages, including Russian, Kazakh, Belarusian, and Ukrainian. The results can be saved as plain text or in HTML (hOCR), ALTO (XML), PDF, and TSV formats. The system was originally created between 1985-1995 in Hewlett Packard's laboratory, and in 2005 the code was opened under the Apache license, subsequently developed with the participation of Google employees. The project's source texts are distributed under the Apache 2.0 license. <\/p>\n
Tesseract includes a command-line utility and the libtesseract library for integrating text recognition features into other applications. Notable third-party GUI interfaces that support Tesseract include gImageReader, VietOCR, and YAGF. Two recognition engines are offered: a classic engine that recognizes text at the level of individual character templates, and a new engine based on applying a machine learning system using a recurrent neural network (LSTM) optimized for recognizing entire lines, significantly increasing accuracy. Pre-trained models are available for 123 languages. Performance optimization modules using OpenMP and SIMD instructions like AVX2, AVX, AVX512F, NEON, or SSE4.1 are also provided. <\/p>\n
Key Improvements: <\/p>\n
\n
Support for RISC-V V vector extensions has been added, on which assembly optimizations for RISC-V processor systems have been prepared.\n
When writing the result in hOCR format, the parameters ocrp_dir and ocrp_lang are set in the created file.\n
The code for determining available language models has been modernized.\n
The code for generating hOCR format files has been improved, and file name transformations on Windows platforms have been removed.\n
Character values are allowed in the options «—oem» and «—psm».\n
In the code, the functions access and _access have been replaced with the method std::filesystem::exists(). The tprintf functions have been replaced with the use of the tesserr stream.\n
Support for the machine learning platform Tensorflow has been removed, which was once implemented but never utilized for executing AI recognition models.\n
The installer for Windows platforms has been improved.\n
The googletest submodule has been updated to version 1.15.2. <\/ul>\n