Hello?
Currently, we are developing in the following environment.
- FRE version: 12.4.7.0
- OS version: Windows 10 (x64)
- Development environment: JDK 8
There is a problem in text block analysis in some documents and it is improved through self-developed post-processing.
In the case of the current algorithm, it is necessary to detect the separators.
Even if the engine parameter is changed, the separators are not detected.
we are using predefined profile (DocumentConversion_Accuracy) and the following profile
[PrepareImageMode]
KeepOriginalCoordinatesInfo = true
[PagePreprocessingParams]
CorrectOrientation = true
[PageAnalysisParams]
EnableExhaustiveAnalysisMode = true
DetectTables = false
DetectSeparators = true
DetectPictures = false
DetectVectorGraphics = false
[ObjectsExtractionParams]
DetectPorousText = true
DetectTextOnPictures = true
[RecognizerParams]
TextLanguage = English
LowResolutionMode = true
You can check out the image we used for the test in the link here.
Table detection is not considered because it cannot be applied due to various problems when testing in a real environment.
How can I detect separators in the image by changing the parameter?
Thank you!