Tesseract

2022: Tesseract 5.2 with improved assembly scenarios for Autotools, CMake

The release of the optical Tesseract 5.2 system text recognition has been published, which supports the recognition of UTF-8 characters and texts in more than 100 languages, including,, and Russian Kazakh. Belarusian Ukrainian This became known on July 7, 2022. The result can be saved both in plain text and in formats (hOCR HTML), ALTO (XML), and TSV. PDF Initially, the system was created in 1985-1995 in the company's laboratory, in 2005 the Hewlett Packard code was opened under license Apache and subsequently developed with the participation of company employees. The source Google code project is distributed under the Apache 2.0 license.

Tesseract includes a console utility and libtesseract library for embedding text recognition functions in other applications. Of the third-party GUI interfaces supporting Tesseract, gImageReader, VietOCR and YAGF can be noted. Two recognition engines are offered: a classic one that recognizes text at the level of individual character patterns, and a new one based on the use of a machine learning system based on the LSTM recurrent neural network, optimized for recognizing entire lines and allowing for a significant increase in accuracy. Ready-made trained models are published for 123 languages. To optimize performance, modules are offered that use OpenMP and SIMD instructions AVX2, AVX, AVX512F, NEON or SSE4.1.

Key improvements in Tesseract 5.2:

• Added optimizations implemented using Intel AVX512F instructions.
• The C API implements a function for initializing tesseract with loading a machine learning model from memory.
• A invert_threshold parameter has been added that determines the level of inversion of text strings. The default value is 0.7. To turn off inverting, set the value to 0.
• Processing of very large documents on 32-bit hosts has been established.
• A transition was made from using the functions std:: regex to std:: string.
• Improved assembly scripts for Autotools, CMake and continuous integration systems ^[1].

2017: Announcement

At the end of August 2017, the consulting company Ernst & Young (EY) announced an automotive blockchain system that is designed to simplify the car sharing business. The new development was called Tesseract.

With its help, companies or a group of individuals will be able to make it easier to share cars and organize access to cars and trucks.

It is assumed that Tesseract will become a single platform to which automakers, mobile service developers, transport companies, infrastructure solution providers, etc. will be connected.

As EY partner John Symlett told Reuters news agency, Tesseract will allow sharing and ownership of large fleets, for example, a group of cars parked on a special site. It will be possible to access the car using the smartphone application.

The publication notes that the concept of vehicle sharing is not new to the market, but the introduction of blockchain projects for registering car rights, maintaining documentation on the use of machines and distributing insurance costs and other needs is in its infancy.

Testing of the Tesseract system will begin in the fourth quarter of 2017. To do this, EY invited a partner whose name was not disclosed. Detailed information about the technology has not been disclosed.^[2]

Tesseract will enable innovation, commercialization and scaling of new mobility models and revenue sources. Thanks to the platform, cars will be used more often, but at the same time costs will decrease, and the transport world will become more environmentally friendly, "said Randy Miller, head of the global transport department at EY.

Notes