Change Log

Released in August 1, 2024

Changes since version 3.5.6.0:

• Added wrapper for Python 3.12.

Released in June 3, 2024

Changes since version 3.5.5.0:

• Letter 'Q' is now correctly recognized in license plates from Azerbaijan.

Released in July 3, 2023

Changes since version 3.5.4.0:

• Added support for license plates from Andorra.
• Added support for the latest 'AA AA NNN' license plate format introduced in 2022 for Hungary.

Released in May 25, 2023

Changes since version 3.5.3.0:

• Added wrappers for Python 3.10 and Python 3.11.
• Added support for license plates from Turkmenistan.
• Added support for Bharat pan-Indian license plates.

Released in March 15, 2023

Changes since version 3.5.2.0:

• Added missing light on dark license plate templates for India.
• Fixed the Mercosur motorcycle license plate templates from Brazil.

Released in February 28, 2023

Changes since version 3.5.1.0:

• Added support for license plates from Mozambique.
• Added support for vintage, temporary, and diplomatic license plates from Spain.

Released in January 5, 2023

Changes since version 3.5.0.0:

• Added new overloads to 'IProcessor.analyze' that take a JPEG, PNG or TIFF encoded image from a buffer in memory or a 'Stream' object.
• Floating point exceptions are now disabled during execution of all variants of the 'IProcessor.analyze' call.
• Project migrated to Visual Studio 2022.
• Dependency libraries have been upgraded to their latest version.
• Dropped support for Windows 7 and Windows 8.

Released in November 21, 2022

Changes since version 3.4.9.0:

• Added wrappers for Python 3.8 and Python 3.9.
• The 'SimpleLPR3.dll' assembly now targets .NET Standard 2.0 for better interop with the .NET Framework, .NET Core, and .NET.

Released in September 25, 2022

Changes since version 3.4.8.0:

• Added support for license plates from Zimbabwe.

Released in August 4, 2022

Changes since version 3.4.7.0:

• Added support for license plates from Tanzania.

Changes since version 3.4.6.0:

• Added support for license plates from Pakistan.

Changes since version 3.4.5.0:

• Improved recognition of Turkish square plates.

Changes since version 3.4.4.0:

• Added support for license plates from the Philippines.

Changes since version 3.4.3.0:

• Added support for pre-2001 'AAA NNNA' license plates from the United Kingdom.

Changes since version 3.4.2.0:

• Added support for Kemler codes.

Changes since version 3.4.1.0:

• Fixed a rare 'Clipped candidate ROI too small' exception that could result in a deadlock condition when the 'plate region detection' or 'crop to plate region' modes were enabled in 'IProcessor'.

Changes since version 3.4.0.0:

• GPU functionality migrated to CUDA 11.3.
• Lowered working memory requirements.
• Added support for two-line license plates from Jordan featuring a single digit in the upper line.

Changes since version 3.3.5.0:

• Added a new crop to plate region feature to IProcessor that constrains the segmentation of the license plate text to the region extracted with the license plate region detection functionality. This can help in cases where there is no gap between the text elements and the license plate frame.
• Performance improvement when employing multiple IProcessor instances in GPU mode.

Changes since version 3.3.4.0:

• Added support for license plates from Belarus.
• Added support for two line, diplomatic, and military license plates from Bosnia and Herzegovina.
• Speed improvement in the license plate bounding quadrilateral regression.

Changes since version 3.3.3.0:

• Better resolution of '1'/'I' and '0'/'O' ambiguities on recognized license plates not assigned to a specific country.
• Brazilian Mercosul license plates with 'AAAN{I}NN' format are no longer misinterpreded as old-format 'AAA-N{1}N'.
• Added support for two line license plates from Cuba.
• Letter 'Q' is now correctly recognized in old-format license plates from Ecuador.

Changes since version 3.3.2.0:

• OCR improvements in the case of merged characters.
• Added Visual Studio 2019 solution and project files.
• Added support for non-passenger license plate types from Colombia.

Changes since version 3.3.1.0:

• Added missing license plates of foreigners, stateless persons, and foreign legal entities in Kazakhstan.
• Fixed yet another very rare 0xC0000005 access violation condition.

Changes since version 3.3.0.0:

• Performance improvements on both CPU and GPU pipelines.
• Fixed issue where two consecutive '1' or 'I' characters were sometimes recognized as a single character.

Changes since version 3.2.6.0:

• Implemented new functionality to check that candidates actually correspond to license plates rather than to incidental scene text. When this feature is enabled, a score quantifying the confidence that the detected text belongs to a license plate, along with the four vertices of the quadrilateral that encloses the license plate, is returned for each candidate.
• Fixed a very rare data race condition that could result in 0xC0000005 access violation exceptions, especially on images containing a lot of text.

Changes since version 3.2.5.0:

• Added missing light on dark 'NN{M} NNN.NN' license plate template for Vietnam.
• Fixed rare NaN condition that could lead to a 0xC0000005 access violation exception, mainly on the x86 version.

Changes since version 3.2.4.0:

• Added support for license plates from Kyrgyzstan.

Changes since version 3.2.3.0:

• Added new 'AA-NN-AA' template for Portugal.
• Added support for diplomatic and National Republican Guard license plates from Portugal.

Changes since version 3.2.2.0:

• Added support for license plates from Macau.
• Added new 'AAANNA' template for Sweden.

Changes since version 3.2.1.3:

• Added missing light on dark 'NN AAA NNN' and 'NN AAA NNNN' templates for Turkey
• Added support for dark on light military license plates from Vietnam.

Changes since version 3.2.0.0:

• GPU functionality migrated to CUDA 10.1

Changes since version 3.1.0.0:

• Added support for x86.
• The requirement of an AVX instruction set enabled CPU has been removed.
• Improvements in performance and memory consumption.