Working with Profiles
ABBYY FineReader Engine supports numerous parameters that let you fine-tune the Engine. In most cases, you will not need to set all the values yourself: the properties of new objects are set to reasonable defaults on creation, and for main usage scenarios we provide well-tested predefined profiles.
If your scenario is more specific and you have determined the optimal settings for the best processing speed and quality, create a profile of your own. All parameters for image preprocessing, analysis, recognition, synthesis, and export can be changed via a profile. For example, if your application exports recognition results to TXT (plain text without formatting), many layout-related properties are irrelevant and can be switched off.
After loading a profile with the help of the LoadPredefinedProfile/LoadProfile methods of the Engine object, the properties of newly created objects will be set to default values specified in the profile.
Predefined profiles
ABBYY FineReader Engine provides a set of predefined profiles designed for the main usage scenarios. Most of the profiles come in two forms: optimized for the best quality of the resulting document or optimized for the highest processing speed. For the complete specification of all predefined profiles, see Predefined Profiles Specification.
| Profile name | Description |
|---|---|
| DataExtraction | Suitable for extracting data from a document and presenting it in a structured format. All objects, including tables, pictures, checkmarks, and handwriting are saved together with their location and attributes. This profile is best for further processing the extracted data or integrating with other applications. |
| DocumentConversion_Accuracy |
Suitable for converting documents into an editable format (for example, RTF, DOCX). The settings have been optimized for accuracy:
|
| DocumentConversion_Normal |
Suitable for converting documents into an editable format (for example, RTF, DOCX). The settings have been optimized for processing speed:
|
| DocumentArchiving_Accuracy |
Suitable for creating an electronic archive (converting to PDF, PDF/A, PDF, and PDF/A with MRC). The settings have been optimized for accuracy:
|
| DocumentArchiving_Speed |
Suitable for creating an electronic archive (converting to PDF, PDF/A, PDF, and PDF/A with MRC). The settings have been optimized for processing speed:
|
| TextExtraction_Accuracy |
Suitable for extracting text from a document. The settings have been optimized for accuracy:
The text is extracted in the same order in which a human would read it, which ensures consistency of data when used for training and processing with natural language processing (NLP) engines.
|
| TextExtraction_Speed |
Suitable for extracting text from a document. The settings have been optimized for processing speed:
The text is extracted in the same order in which a human would read it, which ensures consistency of data when used for training and processing with natural language processing (NLP) engines.
|
| FieldLevelRecognition | Suitable for recognizing short text fragments. Currently, this profile has default settings. |
| BarcodeRecognition_Accuracy |
Suitable for barcode extraction. Extracts only barcodes (texts, pictures, or tables are not detected). The settings have been optimized for accuracy. For compatibility purposes, you can also access this profile under the BarcodeRecognition name.
|
| BarcodeRecognition_Speed |
Suitable for barcode extraction. Extracts only barcodes (texts, pictures, or tables are not detected). The settings have been optimized for processing speed.
|
| HighCompressedImageOnlyPdf |
Suitable for creating high-compressed PDF files that contain entire documents saved as pictures. The following settings are used:
|
| BusinessCardsProcessing |
Suitable for recognizing business cards. The following settings are used:
|
| MachineReadableZone |
Suitable for extracting data from a machine-readable zone (MRZ). The following settings are used:
|
| EngineeringDrawingsProcessing |
Suitable for recognizing technical drawings. It takes into account the large size and complexity of engineering diagrams, as well as the possibility of different text orientations within the image. The profile is intended for converting such images into searchable PDF format. The following settings are used:
|
| Default | Sets all the processing parameters to the default values. |
The settings provided with these profiles can be loaded using the LoadPredefinedProfile method of the Engine object. After the profile is loaded, newly created objects will have the new default values specified in the profile.
Notes:
- To determine the set of resource files necessary for your application to function with the help of the FREngineDistribution.csv file, consult the page corresponding to the scenario you have chosen.
User profiles
You can also create a custom user profile. The syntax of a profile file is similar to that of *.ini files. Comments can be added by starting a line with a semicolon.
The sections contain the names of the objects in square brackets, and the keys contain the properties with their new values. The special section called UserData can contain any user-defined keys. The properties of different types can be written using the following formats:
- Boolean property values are represented by strings: true or false.
- Enumeration property values are represented by constant names.
- String property values are written out without quotes.
For example:
[PrepareImageMode] RasterizeFreeText = false [PDFExportParams] TextExportMode = PEM_ImageOnText ; this is a comment [RecognizerParams] TextLanguage = English,Russian [TextExportParams] TabSeparator = \t
The LoadProfile method of the Engine object allows you to load a user profile file. After the user profile is loaded, newly created objects will have the properties set to values specified in the profile. Loading parameters from a profile is similar to setting the corresponding properties in the program code, but it simplifies the application logic. If you call LoadProfile with an empty string as input, the standard default values will be used.
The correctness of the new values of the properties and their conformity to the license are checked when creating the corresponding object.
A profile file can be used to re-specify all the properties of the following objects:
1 Note that the UserPatternsFile property of the RecognizerParams object accepts only the full path to the pattern file, with the backslash symbol "\" as a separator. The path value shouldn't be enclosed in quotes. For example:
[RecognizerParams]
UserPatternsFile = C:\folder\file.ptn
2 To set the properties of the PictureExportParams or PaperSizeParams objects, specify the parameters directly in the section of the export parameter object (not in the PictureExportParams or PaperSizeParams section). This will allow you to use different settings for different export formats. For example, to specify gray picture format for RTF files:
[RTFExportParams]
GrayPictureFormats = GPF_Png
3 To set the properties of the DocumentContentInfoWritingParams object, specify the parameters directly in the section of its parent object. For PDF format, it is the PDFExportFeatures object; for other formats, it is the corresponding export parameter object. Thus you can specify different content info settings for different export formats. For example, if you do not want to write document author into output PDF files, insert the following lines into the profile:
[PDFExportFeatures]
WriteAuthor = false
4 To set the properties of the PageMargins object, specify the parameters directly in the section of its parent object. Note that the UseCustomPageMargins property set to TRUE must be specified before the values of page margins:
[RTFExportParams]
UseCustomPageMargins = true
PageMargins.Left = 5000
PageMargins.Right = 5000
PageMargins.Top = 5000
PageMargins.Bottom = 5000
Using both predefined and user profiles
One predefined profile and one user profile can be loaded simultaneously. A user profile has priority over a predefined profile, that is, if the user profile sets the same parameter as the predefined profile, the value from the user profile will be used.
If you load one more predefined profile, this new profile replaces the previous predefined profile. Similarly, a new user profile replaces the previous user profile. Note that loading a profile cleans the current recognition session (equivalent to calling the IEngine::CleanRecognizerSession method).
See also
Tuning Parameters of Preprocessing, Analysis, Recognition, and Synthesis
07.11.2025 12:48:30