Project Description

OCR in .NET.

Puma.NET is a wrapper library for Cognitive Technologies CuneiFrom recognition engine that makes it easy to incorporate OCR functionality in any .NET Framework 2.0 (or higher) application. The API is provided through a number of simple classes. High performance and accurate results of recognition can be achieved via a couple of lines of code.
Reference documentation and sample Windows Forms application is available making it easier getting started developing with Puma.NET.

The OCR capabilities:
1. Omni Font technology – recognition of almost all printed fonts
2. 27 languages supported (English, German, Croatian, Polish, Danish, Portuguese, Dutch, Digits, Czech, French, Romanian, Hungarian, Bulgarian, Slovenian, Lettish, Lithuanian, Estonian, Turkish, Russian, Swedish, Spanish, Italian, Russian-English, Ukrainian, Serbian)
3. Spell checking and automatic corrections
4. Detecting font formatting (sizes, italics, boldness, underline, sans, courier)
5. Fragmentation and preserving document structure: paragraphs, spaces, images, tables etc.
6. Improved recognition of skewed and speckled images, special modes for dot-matrix and fax papers’ scans
7. Input image formats: BMP, GIF, EXIG, JPG, PNG and TIFF
8. Output formats: TXT, RTF, HTML

The SDK is available under BSD license. You may either download and run a setup package that will extract necessary assemblies, documentation, register a COM server or you may download the entire source codes.

What Puma.NET can do and what it can't

Puma.NET can do all that is described above and it's also necessary to add that...
1. It can load a single raster image, recognize it store results in a file (single page recognition)
2. Get recognition results in a string

Puma.NET can't:
1. Load multiple images and save recognition results in a single file (recognize multiple pages)
2. Provide DOM for the recognized image: that's when you want to implement two split windows with recognized text in one and highlighted boxes for the text on image in the other.

Recognition results

CuneiForm received the mark 'Editors Choice' by PC Expert Magazine. The CuneiForm recognition engine was repeatedly acknowledged the to be best OCR according to Russian Academy of Sciences conducted tests.


1. Install package (binaries, documentation, sample application): PumaNET.msi (25 Mb)

2. Source code: (28 Mb)
3. Getting started: Getting started.doc, Getting started.pdf
4. Reference documentation: PumaNET.chm

Last edited Sep 14, 2011 at 2:55 PM by MaximSaplin, version 15